token-optimizer-mcp

---

How it compares to Alternative A

Providers cache the prompt prefix: cached tokens re-read at 0.1x, rewritten ones bill at 1.25x. Most compressors optimise bytes removed and ignore that multiplier. This one optimises the bill.

Measured against their real implementation, on their own fixtures. Not a reimplementation and not our fixtures: their harness compresses the payload, dumps it, their resolver redeems their own markers, and ours is handed the identical bytes. Their column is the best of the nine configurations that capture holds, row by row.

Two of our arms are in the table, and they answer different questions. ours is the block engine with nowhere to move a block, which is how the MCP tools and the hooks run: everything it removes, it removes by compressing, and nothing leaves the request. ours, proxy is the proxy a Claude Code session runs through, which also moves a block it could not compress out to a file and leaves a marker naming the path -- the same kind of trade their content-cache reference makes. ours, preset is the block engine with that move switched on at spillWholeBlockBelow: 0.9.

workload	payload	theirs	ours	ours, proxy	ours, preset	zero-turn ids
agentic-conversation	244,908	37.3% / 33.4%	42.9% / 30.8%	52.4% / 49.0%	100.0% / 100.0%	2099 / 2099 / 0
agent-loop	170,938	42.2% / 54.3%	69.3% / 59.1%	92.0% / 94.1%	93.6% / 94.3%	185 / 185 / 52
agent-loop-logs	327,500	51.5% / 53.0%	61.9% / 58.1%	88.7% / 91.5%	96.2% / 97.0%	2034 / 2034 / 81
api-responses	49,127	88.2% / 89.3%	66.6% / 59.1%	96.4% / 96.8%	96.6% / 96.8%	699 / 98 / 23
browser-session	781,774	21.7% / 13.1%	93.7% / 26.5%	55.5% / 90.2%	100.0% / 99.9%	334 / 334 / 0
code-search	131,264	47.4% / 59.7%	81.8% / 70.8%	97.7% / 97.9%	95.7% / 96.1%	201 / 201 / 43
database-rows	45,929	88.6% / 89.2%	65.6% / 56.5%	92.4% / 93.1%	92.6% / 93.4%	590 / 78 / 49
codebase-exploration	137,961	99.7% / 99.7%	16.0% / 12.7%	99.6% / 99.6%	100.0% / 99.9%	540 / 3 / 0
grep-output	75,219	99.6% / 99.6%	56.1% / 51.9%	99.5% / 99.6%	99.9% / 99.9%	1005 / 3 / 0
log-entries	91,385	80.8% / 83.2%	63.5% / 52.0%	95.4% / 95.7%	95.5% / 96.0%	1180 / 237 / 46
human-authored-json	37,425	30.6% / 30.0%	66.6% / 66.0%	81.6% / 83.7%	96.7% / 96.9%	154 / 154 / 11
issue-triage	109,354	53.1% / 55.2%	88.1% / 85.7%	97.2% / 97.6%	95.1% / 95.7%	59 / 59 / 21
rag-conversation	171,913	0.0% / 0.1%	84.6% / 83.2%	99.1% / 99.4%	100.0% / 100.0%	42 / 42 / 0
raw-build-log	158,057	99.8% / 99.8%	70.6% / 48.2%	99.7% / 99.8%	100.0% / 100.0%	430 / 3 / 0
relevance-probe	49,157	64.1% / 71.0%	82.4% / 77.2%	97.1% / 97.8%	95.8% / 96.6%	225 / 225 / 13
search-results	96,647	95.0% / 95.4%	74.3% / 59.4%	96.5% / 96.5%	96.6% / 96.6%	505 / 62 / 34
repeated-reads	158,285	37.4% / 30.4%	71.7% / 68.5%	99.5% / 99.6%	100.0% / 100.0%	394 / 388 / 0
sre-debugging	312,276	88.4% / 89.9%	67.4% / 62.4%	98.5% / 98.7%	97.1% / 97.5%	3116 / 506 / 82

Recorded 2026-10-17 at b1231a8c, by the command in the record's regenerate field, over Alternative A's own fixtures and harness. That arm runs Alternative A itself, so CI does not re-derive it -- it checks these figures and this provenance against the record instead.

Over the corpus the proxy removes 82.5% of the characters and 91.5% of the tokens; their best configuration removes 51.5% and 61.4%. It is cheaper than their best arm on 18 of 18 rows when nothing is ever fetched back, which is the end of the range their markers are built for, and on 17 of 18 when every moved block is fetched back. The block engine alone removes 72.0% and 55.9%: more of the characters than theirs, fewer of the tokens.

Where they still win, and why. On codebase-exploration, grep-output, raw-build-log, sre-debugging and the four raw-data rows their best arm takes 83.2-99.8% off the tokens and the block engine 12.7-62.4%. Their number there is a content-cache reference: the block is not made smaller, it is taken out of the request, put in a store, and replaced by a <<ccr:hash,kind,size>> marker. An engine with nowhere to put a block cannot match that, so on those rows the ours column loses on reduction, and on cost when nothing is fetched back. The proxy makes the same move: it is ahead of them on five of those eight rows, level on grep-output and raw-build-log, and a tenth of a point behind on codebase-exploration. The one row where the proxy costs more than they do is agentic-conversation, and only when every block it moved is fetched back: they move none of that conversation out, so they cost the same at either end, while the proxy moves part of it and pays a retrieval for each part it wants back.

The other column is a turn. A marker is not the content: to read what it stands for, the agent spends a request. Of the 13,792 retention units planted in the corpus, the block engine keeps 5,223 in the text and 8,569 more rebuildable from the text alone, so every one of them is there with no extra turn; theirs keeps 6,711 in the text and 7,075 behind its store, one retrieval call away. 0 are unrecoverable on our side and 6 on theirs. Their best arms force 17 round trips across the corpus to get their content back; the block engine forces 0. The proxy, like theirs, trades a turn for every block it moves, and that turn is what the fetch-rate end of the cost model prices.

What their marker costs is priced, not just its size. Their marker is only redeemable because their proxy also sends a alternative_a_retrieve tool with the request -- on by default (ccr_inject_tool), and on every later request once a session has used it -- so every row where their best arm leaves a marker is charged for that tool's definition as well, recorded from their own package by bench/compression/alt-a/ccr-tool.py. The turn that first adds the tool also rewrites the whole cached prefix behind it; that is not charged, which errs in their favour. Ours needs no tool: our marker names a path the agent opens with the Read it already has, and the path is priced at the length the proxy really writes (<home>/.tok/<pid>/<15 digits>, src/compress/spill-path.ts) rather than a shorter invented one.

browser-session is mostly the fixture. This payload holds four images, two of them distinct, and the base64 in them is generated rather than photographic: one distinct image alone folds from 160,032 characters to 5,572, which no real PNG would do. What carries over to a real session is the duplication -- half the image bytes are a second copy of an image already in the request, which is 44.3% of the whole payload. The run the marker names is still in the output above it, so the reader rebuilds it without asking for anything, which is why the characters fall much further than the tokens: the image tokens are counted from pixels on both arms either way.

Reproduce the whole table:

python bench/compression/alt-a/run-theirs.py <alt-a-clone> <out-dir>

python bench/compression/alt-a/resolve-theirs.py <alt-a-clone> <out-dir>
node bench/compression/head-to-head.mjs <out-dir>

Against their published design

A second arm reimplements their published design from their own benchmark generator's definitions — opaque hash markers, history compressed, a retrieval tool and system message injected — so the four workloads with a published comparator can be checked without their clone. node bench/compression/proof.mjs:

workload	payload	theirs	ours	verdict
code-search	17765	92.1%	97.8%	ours
sre-debugging	65694	92.2%	98.6%	ours
issue-triage	54174	72.8%	97.5%	ours
codebase-exploration	78502	47.4%	54.7%	ours

Four workloads, because those are the four this arm has a published comparator for. The head-to-head harness scores eighteen; the other fourteen are not in this arm, and are scored against their real implementation in the table above.

codebase-exploration was theirs and is now ours, by 7.3 points. An earlier version of this table claimed 61.3% for us on that row, then parity, then a published loss at 46.0% against their 47.4%. The first two were figures from a branch rather than from here; the third was true on this tree when it was written; four log engine commits then moved it to 54.2%, masking an interior hashbang so a concatenated block still parses moved it to 64.9%, and authenticating every marker with a six-character stamp took 0.2 of a point back, and naming a spilled file by a digest of its contents, the way the proxy sink really names it, instead of by its position in a counter, took another half point, to reach 64.2%. Then the currency stopped being an estimate: every figure in the table above is now the count Anthropic's own tokenizer returns for the text, recorded into a committed fixture so this stays offline, and under it the row reads 53.7%, and pricing the path the proxy's sink really writes -- content-addressed and decimal, where this arm had invented a 32-digit hex name -- moved it to 54.7%. That is the largest move in this history and the only one that is not about the engine at all. Dividing characters by four had been charging the control's hex markers too much and our own stamp far too little -- the stamp was six characters of a vowelless base-31 alphabet then, and a random string over a large alphabet is close to the worst case a BPE tokenizer has, so it cost a measured 5.52 tokens where chars/4 billed 2. Those three are the only figures here that moved the wrong way, and none of them is a regression: each is the harness charging us for something it had been leaving out. Being able to see that cost is what paid for the next move: the stamp is now nine decimal digits, which carry slightly more entropy than those six characters did and cost 4 tokens on every draw rather than a mean of 5.52 that ranged from four to seven. Measured against the old stamp on this row -- the same tree, the same fixture, only the stamp pinned back to its base-31 form -- that bought a tenth of a point, and a tenth is the honest size of it: a marker got better than a quarter cheaper, but markers are a small part of what the engine emits, so the headline moves far less than the per-marker saving suggests. A number in prose is a fact about the tree it was measured on, which is why bench/compression/readme-table.check.mjs re-derives every figure in the block above from the harness rather than trusting it.

Against the four comparators in this reimplemented arm: ours on all four. Against their real implementation the answer is the table at the top of this section, and it is a split rather than a sweep: the proxy is cheaper than their best arm on every row when nothing is fetched back and takes more of the tokens overall, while the block engine alone keeps every identifier without a turn and loses on reduction wherever their best arm moves a block out of the request.

The two columns come from different arms of the same engine, and that is the point. v3-history compresses history too and matches them byte for byte; v1-frontier leaves the cached prefix alone and wins the invoice by 11–42%. Maximising bytes removed is available and is not the default, because on a cached prefix it costs money.

The competitor arm is this repository's reimplementation of their published design — opaque hash markers, history compressed, a retrieval tool and system message injected — not their binary, and it is held to the same signed-content guard we hold ourselves to.

On whole sessions

node bench/compression/session-replay.mjs replays real recorded conversations turn by turn and prices each the way a provider does. Four independent sessions, whole request including system prompt and tool schema, against no proxy:

conversation length	ours	cheaper on
10 turns	0.953x	4 of 4
20 turns	0.915x	4 of 4

This is the region nothing else touches: conversation history is ~65% of a live request and the only part that grows every turn. The saving grows with session length, because model reasoning accumulates.

On end-to-end agent tasks

THOL, 16 real tasks against a no-proxy control:

• cheaper on 11 of 16 tasks
• median cost 0.926x, median turns 0.671x
• zero quality cost: mean score 0.994 against control's 0.994, lower on 0 of 16 tasks

One run per task, so the aggregate interval still spans 1.0; the per-task tally and the score parity are the solid parts. Every figure here is regenerated from the committed benchmarks, which ship in this repository.

The 30-second version

Your agent burns most of its context on work it already did: re-reading files that have not changed, dumping a whole file to see three lines, running unbounded searches, and re-deriving conclusions it reached last session and then forgot.

Token Optimizer attacks that on four fronts.

1. It makes the expensive call impossible. Install the plugin and a built-in Read of a 200 KB file is denied, with the refusal naming the cached, diffed replacement. Same for Grep, Glob, Edit, Write, and cat / head / grep -r through the shell. Re-reading a file you already read this session returns only a diff — usually the single biggest win, and one that size-based rules structurally cannot catch. There is no setting to turn on.

2. It remembers what your agent worked out. A per-project knowledge graph accumulates findings, decisions and dead ends as a side effect of working, then feeds them back the moment the agent touches the relevant file. A finding costs ~150 tokens to carry. Re-deriving it costs 5k–50k.

3. It measures itself, in public, and tells you when it is losing. A materialized before/actual-return measurement for MCP progressive disclosure, with later expansions debited from the same net. Modeled graph substitutions and the randomized control arm for downstream graph effects remain separate. Every number is measured, visibly collecting, excluded, or absent.

4. It attributes the traffic. Returned context, optional cost equivalents, and net transport avoided are grouped by operation and MCP handshake identity. Codex, Claude Code, Gemini, and any other connected client get separate rows. Old records without identity remain explicitly unattributed instead of being assigned to whichever agent happens to be open now.

No account and no hosted service. MIT, so it is usable at work.

Nothing is measured about you unless you ask for it. Anonymous usage data is off by default and takes two separate opt-ins: TOKEN_OPTIMIZER_TELEMETRY=1 aggregates counts into a file on your own disk, and TOKEN_OPTIMIZER_BEACON=1 is what uploads any of it. Setting the first opens no socket. With both set, the events already on disk are sent once per session at start-up -- in a batch, never on the hot path -- and the file is cleared only after the receiver has accepted them. install_doctor prints which of the two is on, how many events are waiting and where they would go; it reports without sending. A package built without the upload key cannot transmit at all, and the doctor says so rather than letting an opt-in look like it is working. DO_NOT_TRACK=1 overrides both, whatever else is set. The version check npm run doctor and install_doctor make is a plain GET to the registry npm installs from, carries nothing about you, and runs only when you ask for a diagnosis -- never during a session. DO_NOT_TRACK=1 or TOKEN_OPTIMIZER_UPDATE_CHECK=0 suppresses it before a socket is opened. An event may only contain numbers and flags -- a string is dropped before it is written, so a path, a prompt or an error message cannot travel even by mistake -- and the machine identifier is a salted hash, never your hostname. What the counts are about is the hook ledger your own project already keeps: how often the graph answered instead of a tool, what each arm of the holdout was delivered and withheld, and whether the saving exceeded the cost -- reduced to integers once every six hours, never the ledger itself, which names your files. Usage is counted as a rollup rather than a row per request: one event per 200 requests, holding the request count, bytes in and out, how often compression paid, how often knowledge was injected, and how many blocks were elided, plus one final event when the proxy stops. That keeps the totals exact -- a row per request would overflow the batch on a busy session and leave you reporting its first five hundred -- at the cost of the per-request distribution, which is not something anyone downstream needs. npm run doctor prints exactly which of these is on and what has been recorded.

What the dashboard proves on a real machine

The screenshots in this README come from the shipped server reading persisted local data, not a design mockup. In the capture above it reports:

• 43,491 net verified MCP transport tokens avoided in the current live proof: 54,037 gross reduction minus a deliberate 10,546-token expansion;
• 486,074,740 historical/tool-reported tokens quarantined, dominated by repository scan volume that never entered model context;
• a live Codex / Claude Code / Gemini stdio smoke against AiDotNet, with each client attributed independently;
• 2,648 graph nodes, 6,527 edges, and 58 findings across 11 local projects;
• more than 1,000 hook runs with zero failures and zero timeouts across six active CLI clients in the selected rolling 24-hour window;
• 6,332 tokens of modeled graph-substitution potential, excluded from the verified headline while the causal graph-reuse study remains Collecting.

The dashboard now reads native CLI usage receipts and prices uncached input, cache reads, cache writes, and output with the exact captured provider, model, route, request-time tier, and versioned official source. Ambiguous model ids stay Not priced instead of receiving a blended guess. API/list-price equivalents are kept separate from provider-reported charges and are never labeled as a subscription invoice. See the token accounting contract.

Quick start

Claude Code — install the plugin, not the bare MCP server. The plugin is what enforces; adding the server alone just gives the model tools it can ignore.

/plugin marketplace add ooples/token-optimizer-mcp
/plugin install token-optimizer@token-optimizer
/reload-plugins

That is the entire installation. All sixteen clients →

Then, whenever you want to know what to do next:

token_audit

One ranked queue: what is costing the most per session, with an optional monthly cost equivalent only after you configure your own effective rate. Each line names how to fix it. Not a dashboard, not six reports — a queue.

---

Subagents and workflows

Agents are where token spend concentrates. A long-running subagent re-sends its whole context on every turn, so every byte a tool adds is paid for again on each later turn. In one measured session, 42 workflow subagents made 4,483 model calls. The six longest made 218–348 calls each and peaked at 336K–609K tokens of context. Together they spent 4.0M output tokens and 1.0B cache-read tokens. The bill is the re-reading, not the writing.

What works with no setup

The plugin's hooks run for subagents exactly as they do for the main session. Nothing needs to be configured per agent.

• Whole-file reads are outlined, through Read and through the shell. Agents mostly read files with cat, type or Get-Content, rarely with Read. In the measured session that was 2,071 shell reads against 9 Read calls. On the first whole read of a large source file, either kind returns a line-numbered outline of its symbols instead of the file. A second whole read of the same file returns the file itself. A line range is never outlined: sed -n 'A,Bp', head, Get-Content -TotalCount and a paged Read all pass through untouched.
• Each subagent is briefed once. A subagent never sees the SessionStart guidance, so its first tool call carries a four-line briefing instead: load the optimizer tools if they are listed, read ranges rather than whole files, and keep the final report short. The tools are named only conditionally, so a hooks-only install is never told to call something it does not have.
• Advisories are said once. A routing hint such as "use smart_grep" appears at most once per agent for each kind of call, not on every matching command.
• Past findings arrive only when they name what you are running. A finding with no explicit trigger is delivered with a command only when its claim names that command's program and subcommand, such as npx jest or dotnet build, not when the two merely share a word.
• Repeated build and test output is bounded, as it is in the main session.
• Through the proxy, a range you asked for comes back whole. A paged Read or a shell slice is never folded, and a fold inside a whole-file read points at the file's own line numbers.

What to add when you brief an agent

Claude Code gives subagents MCP tools as deferred tools: the schemas are not loaded until the agent asks for them. An agent that never loads them falls back to shell commands. Put this in the agent's prompt:

Before your first file read or search, load the token-optimizer tools:
ToolSearch("smart_read smart_grep smart_glob").
Use smart_read for files (it returns a diff on re-reads), smart_grep for
searching, and smart_glob for finding files. Do not print whole files or search
the tree through the shell. When you need part of a file, read a line range.

The server's own instructions now say the same thing to every client, deferred or not.

Keep agents short

The plugin can only shrink what enters an agent's context. The number of turns and the size of the context it carries are set by how the work is split up:

• One bounded task per agent. Several short agents cost less than one agent that runs for hundreds of turns, because each starts from a small context.
• Give the agent the facts it needs (paths, line numbers, the exact question) instead of asking it to rediscover them.
• Ask for a structured result so the agent stops when it has the answer.

Measuring an agent

Each subagent writes its own transcript next to the session's, under ~/.claude/projects/<project>/<session>/subagents/. The usage block on every assistant row records input_tokens, output_tokens, cache_read_input_tokens and cache_creation_input_tokens.

Count each response once. One model response is written as several assistant rows that share a message.id, and each row repeats the usage recorded so far. Keep the LAST row per message.id, which holds the final usage, and sum those. Summing every row overstates the cost: on the session above it reported 1.8B cache-read tokens instead of 1.0B. Keeping the first row instead understates output, at 0.5M instead of 4.0M.

The number of distinct message.ids is the number of model calls the agent made. Read against its peak context, it shows whether the agent should have been split.

---

The knowledge graph — the part nothing else has

Every agent session ends the same way: the reasoning evaporates. The next session re-derives it, at full price, forever.

This builds a living per-project graph — nodes for files, symbols, tasks and findings; edges for derived_from, contains, supersedes, contradicts, related — and it fills itself in from real work. No ingestion job, no embedding model, no rebuild step, no query to formulate.

you touch  src/auth.ts
           │
           ├─ verify() compares exp against the LOCAL clock          (finding, 0.9)
           ├─ per-host retry budgets; global was rejected — deadlock (decision)
           ├─ ! the skew fix was reverted once already               (dead end)
           └─ [git] 47 changes in 90d, last three: "fix token expiry",
                    "revert skew fix", "fix token expiry again"

None of that is in your repository. It exists only because an agent once burned tokens finding it out — and every other tool throws it away at the end of the session.

What a default install actually produces. The structural graph — files, symbols, tasks, and the edges between them — is captured from ordinary tool traffic with no configuration at all. Findings are produced two ways. At session end, derive reads evidence already on disk (command outcomes and exit codes, red-to-green transitions, corrections, re-read churn) and writes findings from it: no model call, no credential, nothing sent anywhere. And the active model records durable conclusions itself through wiki_write.

The model-based semantic harvest is the third path, and the only one that needs something you do not already have. It is not opt-in — TOKEN_OPTIMIZER_HARVEST=0 turns it off — but its real gate is a credential: with none it reports off:no-key, which is the state on CI, corporate machines, and subscription-only logins. Point TOKEN_OPTIMIZER_HARVEST_ENDPOINT at a local model and it runs free and private, with nothing leaving the machine. npx token-optimizer-doctor states which of these is live.

Why this is not RAG

Classic RAG	This
Retrieves evidence; the model re-derives meaning each time	Retrieves verdicts — the reasoning already happened
Index built by a batch ingestion job	Accretes from real agent traffic — coverage follows attention
Similarity search	Traversal — this symbol and its callers
Model must formulate a query	Fires when the model reaches for a file
Staleness invisible; serves rotted chunks confidently	Staleness computed from content hashes, served with the invalidating diff
Returns only what is in the documents	Returns dead ends, which exist nowhere in your source tree

Traversal plus lexical search: deterministic, instant, explainable, and it works offline.

The zero-turn refusal

A plain deny costs a full turn: the model calls Read, is refused, re-plans, calls smart_read. But at refusal time we already hold the file and the snapshot the graph stored — so the refusal carries the answer inside it. Nothing to re-plan, no second call. Turn cost drops from one to zero.

And when the graph already holds the verdict a tool output would support, the output never enters context at all. Not compressed. Absent.

---

The dashboard

npm install
npm run build
npm run dashboard      # http://localhost:3100

The overview answers the questions a token optimizer should answer first:

1. How much MCP context did it avoid? The headline is gross materialized payload reduction minus every later linked expansion. Graph estimates are intentionally separate.
2. How much context still reached the agents? Every successful current MCP result records its actual returned text, even when no valid before-state exists. That row is context-accounted but savings-unmeasured.
3. Which agent and action spent it? The client ledger and action table show operations, returned context, optional cost equivalent, and net tokens avoided. Lifecycle-only clients say Not measured; no zero is invented.
4. Did remembering cost more than it saved? Delivery and semantic-harvest tokens are charged to the graph. A causal benefit is added only after the treated/holdout evidence gate passes.

Walkthrough: get useful data, not an empty dashboard

1. Install the MCP server and the native adapter for your CLI. The MCP handshake provides per-client accounting; native lifecycle hooks provide automatic routing, capture, health, and delivery where the client protocol permits it.
2. Use smart_read, smart_grep, smart_glob, smart_edit, or any other MCP operation normally. Every successful result records returned context; tools with a comparable materialized before-state also record a gross reduction; later expand calls debit that reduction.
3. Let the active model record durable conclusions with wiki_write. Before a new agent re-derives work, call wiki_read for the project or the files it is about to touch. Native clients can also deliver matching knowledge automatically. Use wiki_query to read the graph directly — one finding by key, a ranked BM25 search over claims, a node with its neighbours, or the graph's own audit — which is how a subagent that never sees the SessionStart briefing reaches what previous sessions established.
4. Open http://localhost:3100. Use Overview for combined accounting and What it knows for capture health, graph exploration, audits, and causal evidence.
5. To register existing local repositories without reading their source, run npm run projects:discover -- /absolute/path/to/repos. This makes coverage gaps explicit; it does not fabricate findings.

For maintainers, this live smoke exercises the shipped stdio transport and creates separately attributed rows without seeding the analytics database:

npm run dashboard:attribution-smoke -- /absolute/path/to/project /absolute/path/to/large-file
npm run dashboard:verify-live -- http://localhost:3100

The health panel is deliberately operational rather than a raw text dump. Each client has activity, runtime failures/timeouts, policy outcomes, and observed surface coverage. Diagnostics keep no prompts, commands, paths, or tool output.

Modeled substitution potential and causal graph effects are different claims. The first is a full-file counterfactual and is never promoted to the verified MCP headline. The second asks whether delivered knowledge prevented later reads; it uses a control arm and remains Collecting until there are at least 20 treated file touches and 5 holdouts with valid downstream joins.

Drag to orbit, scroll to zoom, click a node for provenance, or switch to the bounded one-hop focus view. The default All known projects scope pools local graphs through opaque project IDs; filesystem paths never reach the browser.

Audit tab. Contradictions, stale findings, and low-confidence claims remain reviewable instead of silently becoming model truth.

Evidence console. Client/model/task cohorts, matched effects with 95% intervals, live outcome joins, harm feedback, and capability tiers for all 16 clients. Release and superiority claims fail closed while evidence is missing.

One-click Markdown export. The accumulated graph becomes documentation you can inspect, edit, and commit.

Server-side by design: the browser asks for a neighbourhood, a search result or a page. A mature graph holds thousands of nodes, and shipping it wholesale would make every page load a multi-megabyte download for a view that shows twenty things.

The default All known projects scope combines captured graphs through an opaque machine-local project registry; filesystem paths never reach the browser. Lifecycle hooks register repositories as they are used. To backfill existing local checkouts without reading their source files, run the bounded discovery command against one or more explicit roots:

npm run projects:discover -- /absolute/path/to/repos /absolute/path/to/worktrees

Coverage distinguishes repositories with graph data from known repositories whose capture has not started. The balance cards are backed by persisted events: Memory deliveries counts graph context actually supplied to an agent, Kept back for comparison counts randomized control touches, and Cost of remembering combines delivered-context tokens with measured semantic-write payload cost. Reading avoided stays Collecting or Not measured until at least 20 treated file touches and 5 holdouts exist with a downstream join; the dashboard does not manufacture a savings estimate from missing data.

See the cross-client capability contract and the live evaluation suite.

Cross-client lifecycle diagnostics

Every native hook writes the same bounded JSONL lifecycle record, including Claude Code's custom router and compaction paths. Records carry the client and plugin versions, event, hashed session/turn correlation, latency, outcome, input/output byte counts, and response key shape. They deliberately retain no prompt, command, tool output, file content, or raw working-directory path. The fields include OpenTelemetry log severity and resource semantics so the local files can be collected without inventing a second schema.

npm run diagnostics                         # last 24 hours, summary-first JSON
npm run diagnostics -- --hours 72 --output hook-summary.json
npm run diagnostics -- --include-events --limit 100 --output hook-report.json

Raw event rows are opt-in and capped at 1,000. The default report contains aggregate health and at most twenty recent failures/timeouts, keeping routine troubleshooting output small enough for CLI and model context windows.

The dashboard's Capture health panel shows runs, failures, timeouts and p50/p95 latency by client. Logs rotate at 5 MiB, retain at most 40 files for 14 days, and live under .token-optimizer/logs when a state directory is set (or ~/.token-optimizer/logs otherwise). TOKEN_OPTIMIZER_LOG_DIR, TOKEN_OPTIMIZER_LOG_MAX_BYTES, TOKEN_OPTIMIZER_LOG_MAX_FILES, and TOKEN_OPTIMIZER_LOG_RETENTION_DAYS override those operational defaults.

---

What it does that other optimizers do not

Compression that does not break the cache, or lose the needle

An optional local proxy compresses tool results, search output, logs and conversation history on the way to the model. A hook cannot do this: the PostToolUse output schema is {hookEventName, additionalContext?, classifierContext?} and updatedOutput occurs nowhere in the client, so a hook can add context but never replace a result. The proxy never tries to -- it rewrites the outbound request, where those results already sit as history.

Two things make it different from simply compressing harder.

It never rewrites cached content. A cached prefix bills at 0.1x and a cache write at 1.25x, so compressing history can cut tokens while multiplying the bill. Measured on our own benchmark: compressing behind the cache breakpoint removes more raw tokens on every workload and costs more money on every workload. Compression stops at the frontier, and cache-weighted tokens are reported beside raw ones so the trap is visible rather than inferred.

It keeps the rows that matter. Eliding a long array after the first few rows scored 95.7% on a search payload here and destroyed both the UUID record and the error record planted in it -- the only two rows anyone would have searched for. The engine now keeps every row that departs from the shape, wherever it sits: 92.4% with the needles intact. A gate in the benchmark fails the build if a planted needle disappears, because a size metric alone cannot tell compression from truncation.

Reduction over the content each strategy is permitted to rewrite, on fixtures matching the four workloads Alternative A publishes, is the guarded table earlier in this section -- 97.5%, 97.8%, 98.6% and 54.7% against their 72.8%, 92.1%, 92.2% and 47.4%. It is stated once and checked there rather than restated here, which is how this copy came to claim 98.9%, 98.2%, 92.8% and 61.3% long after the tree had moved.

These are not their corpora, which are unpublished; the code workloads read real files out of this repository and the rest are generated to the shape and scale of their published ones, from their own benchmark generator's definition.

Reduction is not the bill, and the two disagree. Cache changes the arithmetic: a cached prefix is re-read at 0.1x and a rewritten one is charged at 1.25x, so an arm can remove more bytes and still cost more. Run node bench/compression/proof.mjs and that is exactly what happens -- the Alternative A-style arm removes a larger share of every workload and loses on cache-weighted effective tokens on all of them:

workload	raw reduction, ours / theirs	effective tokens, ours / theirs
code search	84.6% / 95.8%	999 / 1,383
SRE debugging	86.2% / 97.1%	1,833 / 2,261
issue triage	89.2% / 95.3%	812 / 1,277
grep output	45.9% / 51.9%	7,146 / 9,514
raw build log	54.7% / 59.4%	14,448 / 16,769
browser session	45.7% / 49.4%	11,398 / 14,685

The competitor arm is this repository's own reimplementation of their published design -- opaque hash markers, history compressed, a retrieval tool and system message injected -- not their binary. It is implemented to win where it can: the raw column is theirs on six of six.

On whole sessions. node bench/compression/session-replay.mjs replays real recorded conversations turn by turn and prices each one the way a provider does, charging the longest byte-identical leading run at 0.1x and everything after at 1.25x. Four independent sessions, whole request including system prompt and tool schema, against no proxy at all:

turns	Alternative A's design	ours (v4-substitute)	cheaper on
10	1.000x	0.953x	4 of 4
20	0.999x	0.915x	4 of 4

The replay uses stored transcripts, which discard reasoning text while retaining signatures. Its savings therefore do not establish that useful reasoning survives compression. It also uses our reimplementation of the competitor's design; these are offline estimates, not results from their shipped proxy.

Live comparison against Alternative A's shipped proxy. The balanced four-arm run recorded in commit f712b4ec used bench/live/ab.sh, rotated each arm through every position, and re-ran pytest independently after each task:

arm	mean weighted input tokens	tests passed
uncompressed control	134,706	4/4
Token Optimizer, previous 1,500-character deferral floor	100,818	4/4
Token Optimizer, zero floor (now the default)	49,053	4/4
Alternative A shipped proxy	48,626	4/4

Weighted input is input + 1.25 * cache_creation + 0.1 * cache_read; it excludes output cost. Both zero-floor Token Optimizer and Alternative A reduced that measure by about 64% versus control. Token Optimizer used about 0.9% more than Alternative A. This small arithmetic repair task does not establish superiority across tasks, compression quality, or total cost. Rotation balances run position, but does not guarantee that shared provider-cache effects disappear.

Codex three-task comparison against Alternative A (2026-09-18). Three synthetic controlled-read tasks, three repetitions, three arms, every arm in every position. All 27 runs returned the correct answer, so this compares cost at equal correctness rather than accuracy. Alternative A ran in --mode token, its compression-first setting, not the cache default that exists to freeze the cache hit rate.

arm	mean input tokens	of which cached	output	requests	agent seconds
uncompressed control	55,632	43,079	200	3.0	19.1
Token Optimizer	42,468	33,052	189	3.0	17.3
Alternative A, --mode token	71,626	51,897	285	4.3	27.3

Input reduction against the control was 22.7% on the log task, 34.3% on the JSON task and 11.5% on the code search; against Alternative A, 54.9%, 10.0% and 42.2%. Every task favoured this proxy on both comparisons, and the JSON column is where Alternative A is strongest and the margin thinnest.

The mechanism is visible in two numbers that move independently. Mean input per REQUEST was 14,156 here against Alternative A's 16,529 and the control's 18,544 -- both proxies compress, and this one compresses harder. Mean requests per RUN was 3.0, the same as the control, against Alternative A's 4.3. A compressor that elides content the agent then has to read back spends a turn recovering it, and on these workloads a turn costs more than the elision saved; that is the failure this project's own deferral floor exists to avoid, and it is what the extra 1.3 requests are.

WHAT THIS DOES NOT SHOW. These are token counts, not dollar costs, and bench/live/report-codex.mjs deliberately declines to price them. Three short synthetic workloads do not establish a universal win. Each task ran in a fresh temporary workspace with no knowledge graph, so graph injection contributed nothing and this measures compression alone. Cached tokens are a subset of the input column, not an additional charge. Alternative A's compression is scored from provider usage rather than byte counts recorded at our own listener.

Reproduce with ALT_A_MODE=token node bench/live/codex.mjs, then node bench/live/report-codex.mjs <evidence-dir>; the figures above were read only after that reporter returned valid: true with balanced: true and no failures. Four earlier campaigns were discarded rather than published: one ran without Alternative A's [proxy] extra so every Alternative A run errored before readiness, one ran Alternative A in its cache default, one was voided by the reporter for an incomplete request capture, and one died on ENOSPC with the host disk full.

The subsequent Claude Code default-versus-aggressive confirmation hit Claude's weekly usage limit and is incomplete. Its missing usage cannot count as a win. Codex validation runs independently through the Responses API.

Latest Codex joint confirmation (2026-09-16). The frozen twelve-family study completed 120 pairs against installed Alternative A. Our proxy passed 119/120 attempts; Alternative A passed 120/120. One local allocation crash left unknown usage. There were 77 strict joint wins (passing, cheaper, and faster), 20 cost-only losses, 12 speed-only losses, nine losses on both measures, one equal-time pair that cost more, and the failed attempt. Across the 119 fully measured pairs only, estimated token cost was 26.9% lower and agent time 30.1% lower. JSON agent time was 5.6% higher. Only logs and refactoring met the simultaneous per-family criteria; the complete-study superiority gate failed. These results do not establish that every task is cheaper or faster. The measured build preserves explicit nulls, retains complete numeric tables, and avoids copying unchanged payloads. See the full results, every loss, and retained crash.

Earlier expanded local Codex confirmation. A preregistered seven-family study ran 70 pairs / 140 attempts against installed Alternative A. Our frozen proxy passed 70/70; Alternative A passed 69/70 with one upstream HTTP 503 and missing usage. The complete-ledger gate therefore failed: superiority was not established. Across the 69 fully measured pairs only, descriptive totals were 25.7% lower estimated cost, 38.9% lower input, and 30.0% lower agent time for our proxy. JSON remained a loss (35.1% higher estimated cost). See the protocol, all attempts, and limitations.

A subsequent rare-string-group fix targets that JSON loss. In three fully measured fresh pairs, input fell 17.7%, estimated cost 13.4%, and agent time 13.6% versus Alternative A, with three requests each. A fourth pair had a retained upstream 503 on our arm; one measured pair still cost more. These are development results, not confirmation of the new build or an all-workload win. See the complete screen and exclusions.

Current amended Codex study. The next fixed 70-pair schedule completed: our proxy passed 69/70 and Alternative A 70/70. Our remaining attempt hit an upstream 503 with unknown usage, so strict superiority was not established. Across the 69 fully measured pairs only, estimated cost was 27.9% lower, input 44.2% lower, and agent time 31.7% lower. Every family's mean cost favored our proxy, including JSON at 27.5% lower, but 13 individual pairs still cost more. The run includes a disclosed repair to a between-case file-hashing failure; the original product and schedule stayed fixed. See the full results, amendment, and retained failure.

Exact search rows, subsequent development. Factoring shared declaration syntax while preserving every identifier and value closed the code-size gap: 37.5% fewer bytes than archived Alternative A on the original 40 losing inputs. Eight fresh local cases also sent fewer bytes. Four balanced live pairs all passed their audits; our estimated token cost was 50.1% lower, input 44.8% lower, and agent time 51.1% lower. This is a small development follow-up, separate from the broader study. See the captured bodies, reconstruction checks, and live evidence.

Cost-loss follow-up (2026-09-16). We investigated all 13 losing pairs and implemented exact ID-prefix factoring plus support for records inside truncated shell envelopes. A fixed follow-up of those 13 cases passed all 26 attempts; 12 pairs favored our proxy, with 33.4% lower total estimated cost. The remaining loss exposed the outer-envelope bug. After fixing it, two balanced controlled pairs both favored our proxy: all four attempts passed, estimated cost was 34.3% lower, and input was 15.9% lower. These development runs retain every original loss; cache and model variation also affect results. See the case-by-case audit, fixes, exact replay, and live evidence.

Local proxy performance (2026-09-16). Against installed Alternative A, 2,160 completed local-upstream requests covered repeated and unique logs, JSON, and code-search output in three rotated arm orders. Our mean latency and process CPU were lower in all six groups. Sampled peak process-tree private memory was 80.0 MiB versus 1,909.9 MiB; this is footprint, not allocation volume. Unique code still forwarded 5.7% more bytes. Alternative A's rate limit was disabled for this throughput measurement; compression settings stayed default. These measurements do not measure model quality or billed cost. See the raw samples, reproduction commands, and limitations.

Live Codex comparison (2026-09-15). Local Codex CLI 0.154.0, gpt-6-astra, and installed Alternative A completed a balanced 27-run campaign. All three arms passed all nine tasks; every initial read was complete, and all provider ledger totals matched Codex's reported usage.

workload	mean input, ours	mean input, Alternative A	reduction vs Alternative A
JSON outlier lookup	39,318	44,043	10.7%
code search	44,427	82,697	46.3%
log diagnosis	39,203	84,308	53.5%

These counts include cached input and extra retrieval turns; they are not raw compression percentages or dollar costs. Output tokens did not improve on every workload. See the Codex evidence and limitations and reproduction instructions. These three synthetic tasks do not establish superiority across all workloads.

A separate 27-run natural-workflow development screen passed all runs and measured 35.5%, 43.5%, and 16.6% less total input than installed Alternative A for retry bug fixes, multi-file refactors, and configuration refreshes. These are small synthetic repositories with natural tool selection. Uncached input and latency have separate results; see the workflow evidence and limitations.

The proxy is enabled by default and binds loopback only.

It is now on without launching anything through us. Until 7.1.0 the only way to be routed was to start the client through our wrapper, so anyone who installed with /plugin and then opened Claude Code from a shortcut, an IDE or the desktop app saved nothing -- and the doctor reported that as a broken installation rather than as a feature that never applied. Installation now writes the endpoint into Claude Code's own settings.json, which is the one thing a session we did not start will read, and a small background proxy serves it. Four rules make that safe to switch on:

• Recorded. Every value written is stored with whatever was there before. token-optimizer-uninstall --apply restores it exactly, and refuses when the value is no longer the one we wrote -- if you change it, it is yours again.
• Never written on hope. The entry appears only after the proxy has actually served the route. A settings file naming a dead port does not degrade politely: the client cannot reach its provider at all.
• Self-healing. The check runs again at every session start, before the first model request, and removes the entry the moment the route cannot be served.
• Hands off what is not ours. A client already pointed at another local proxy is left alone, and a settings file we cannot parse is never rewritten.

Set TOKEN_OPTIMIZER_DEFAULT_ROUTING=0 to keep the wrapper-only behaviour, or TOKEN_OPTIMIZER_PROXY_AUTOSTART=0 if you do not want a local background service at all. Either one also undoes an entry already written.

Managed commands. Global npm installs activate Claude hooks and wrap the client commands you actually have -- claude, codex, gemini, opencode, qwen, crush, droid, cn (Continue), copilot and amp -- in Command Prompt, PowerShell, Bash, and Zsh when lifecycle scripts are enabled. Only commands already on your PATH are wrapped, so installing this does not claim a name for a client you have not installed; TOKEN_OPTIMIZER_MANAGED_CLIENTS takes a comma-separated list (or all) when a client lives somewhere PATH cannot see, and 0 skips shell activation entirely. Windows installs add owned .cmd launchers to the User PATH, so profiles and PowerShell execution-policy changes are not required. For local installs or disabled lifecycle scripts, run token-optimizer-install. Reopen the terminal to load the updated PATH and shell activation. Each managed session starts its own proxy, registers the packaged core MCP tools (including wiki), and shuts the proxy down on exit. token-optimizer-run <client> ... works without shell activation.

Which clients can be routed, and which cannot. Ten have a supported way to redirect model traffic and are listed above. Claude Code and Gemini have exactly one provider each, so they are routed on installation. The rest choose a provider in their own configuration, and are routed once that configuration names an endpoint -- because forwarding on a guess would deliver one provider's credentials to another company. Six clients cannot be routed at all: Cursor, Cline, Windsurf, Kilo, Roo and Zed run the assistant inside the editor process and expose no documented redirect. Everything else -- the optimizer tools, the graph, the hooks -- still applies to them, and the doctor states the limitation instead of counting it as a failure.

token-optimizer-route shows what is running and what configuration has been written on your behalf. Zed is the one client that can be routed and cannot be routed for you -- its provider is a named entry you pick inside the editor, and only you know which endpoint and model it should use:

token-optimizer-route                                                  # what is routed right now
token-optimizer-route zed --upstream https://api.openai.com/v1 --model gpt-4o
token-optimizer-route zed --remove

Zed's provider schema here comes from its published settings documentation and has not been exercised against an installed Zed in this repository. It is written into a file the command can take back out, and nothing is written unless the route is already being served.

Existing custom provider authentication stays in the client, and your own endpoint stays yours: the proxy is inserted in front of whatever you configured, never in place of it. TOKEN_OPTIMIZER_PROXY=0 disables routing and TOKEN_OPTIMIZER_MODE=off disables optimization. npm lifecycle scripts may be blocked, so package installation alone is not proof that activation ran. OpenCode's session plugin uses its resolved configuration and project directory; account credentials and provider files stay in place. Unsupported endpoints and provider modes retain native routing. Claude routing controlled by local managed-policy files or Windows registry policy also stays native. Remote/MDM policy can arrive after launch; the launcher reports observed model traffic, not just listener startup. doctor checks routing configuration. Request-body capture is opt-in via TOKEN_OPTIMIZER_PROXY_CAPTURE; when enabled, it writes plaintext request content to the named directory. Capture queues are bounded to 128 requests and 16 MiB of queued snapshots and metadata. Excess captures are rejected with a warning while requests continue; live audits reject incomplete capture evidence. See the allocation hardening and follow-up results for the reproduced OOM fix, constrained-memory replay, and remaining cost losses.

Your provider key is forwarded in the request headers and is never read, stored or written by the proxy. That is a narrower claim than "nothing sensitive is written", and the difference matters: the proxy does write message content to disk, described next, and it does not inspect that content for secrets. If a secret is in your conversation, it can reach a spill file the same way any other text does.

It will not reach an unencrypted network, though: an upstream that is neither https nor loopback is refused rather than forwarded to, and with no upstream configured the proxy serves only Anthropic's own routes -- so a client for a different provider is told to name its provider rather than having its key sent to the wrong company.

It does write some payload to disk, and you should know exactly when. An elision has to name a way back to what it removed, and content that arrived in a tool result has no file of its own -- so that content is written to a spill file and the marker names its path. This happens only when an engine actually elides something recoverable-by-path: a JSON array tail, a set of function bodies, a passage of prose. Requests below the size floor, requests nothing claims, and every elision that is lossless are all written nowhere.

Spills go under your OS temp directory in token-optimizer-spill/, one file per distinct content, created 0600 (owner read/write only) and named by an HMAC of the content under a salt generated fresh in each proxy process -- so the name discloses nothing and two runs do not collide. They are never read back by us; the agent reads them with the Read tool it already has, which is the whole point of a path instead of a hash.

They are not deleted while the proxy is running, and that is deliberate: a marker whose spill has been swept is exactly the dangling reference this design exists to avoid, and the agent may follow a path many turns after it was written. So retention within one run is bounded only by the distinct content you elide -- there is no quota, and a very long session that elides constantly can accumulate.

When the proxy stops, its spill directory is removed. Each proxy process gets its own directory under token-optimizer-spill/, so stopping one never sweeps another's live paths, and the moment it stops is also the moment no agent can still be following one of them. If a run ends without that cleanup -- a kill signal, a power cut -- the directory is left behind in your OS temp directory and is safe to delete by hand. Running with the proxy off writes no spills at all.

The main switches, and what each one trades

variable	default	what it does
TOKEN_OPTIMIZER_PROXY	on	the compression proxy itself; set 0 to opt out
TOKEN_OPTIMIZER_COMPRESSION	balanced	balanced, aggressive, conservative, lossless
TOKEN_OPTIMIZER_PROXY_KNOWLEDGE	on	put what this project already learned in the cached prefix
TOKEN_OPTIMIZER_PROXY_NET_SAVING	off	never send upstream more bytes than the client gave
TOKEN_OPTIMIZER_PROXY_REPO_PACK	on	one shared snapshot of the repository for every agent on it
TOKEN_OPTIMIZER_PROXY_REPO_PACK_TASK	off	also pack a subagent's task repository when the session runs outside a repo
TOKEN_OPTIMIZER_PROXY_TRIM_SUBAGENTS	on	drop the memory index, skill text and unused tool lists from subagents
TOKEN_OPTIMIZER_PROXY_READONLY_ROUTE	off	down: declared read-only subagents one model tier down (measured to cost more)
TOKEN_OPTIMIZER_PROXY_REPO_PACK_POSITION	first	the pack right after the first cached system block; last appends it
TOKEN_OPTIMIZER_PROXY_PREFETCH	off	fetch the issue or PR a subagent is assigned before its first turn
TOKEN_OPTIMIZER_PROXY_WORKER_TERSE	on	ask subagents for short reports (output is billed 5x input)
TOKEN_OPTIMIZER_PROXY_CLEAR_TOOL_USES	off	server-side clearing of old tool results; measured to cost more (see below)
TOKEN_OPTIMIZER_PROXY_COMPACT	off	server-side compaction past _TRIGGER input tokens (min 50,000)
TOKEN_OPTIMIZER_REPEATS	on	a re-run shell command returns only what changed
TOKEN_OPTIMIZER_BLACKBOARD

Upgrading from 7.6.0

7.7.0 changes these defaults. Each can be switched back, and each was measured before it shipped (see #523, #525 and #526).

New and on by default

• Subagents.
  • A cache marker before each subagent's task (TOKEN_OPTIMIZER_PROXY_SHARE_PREFIX=0 turns it off).
  • A short wait on a sibling's first request so it reads the first one's cache (TOKEN_OPTIMIZER_PROXY_COLD_GATE=0).
  • The memory index, skill text and unused tool lists trimmed from subagents (TOKEN_OPTIMIZER_PROXY_TRIM_SUBAGENTS=0).
  • Subagents asked for short reports (TOKEN_OPTIMIZER_PROXY_WORKER_TERSE=0).
• Repository pack. A shared repository snapshot in the cached prefix, built only for directories under a trusted root (TOKEN_OPTIMIZER_PROXY_REPO_PACK=0).
• Tool output. Bash and Grep results over 6 KB are compacted, with Grep results grouped by file. The full output is stored, and the notice names the file to Read (TOKEN_OPTIMIZER_OUTPUT_COMPACT=0). A re-run shell command returns only what changed (TOKEN_OPTIMIZER_REPEATS=0).
• Siblings and shells.
  • A subagent's briefing lists read-only commands its siblings already ran (TOKEN_OPTIMIZER_BLACKBOARD=0).
  • Agents' shells get quiet tool defaults (TOKEN_OPTIMIZER_QUIET_ENV=0); a value the user sets always wins.
• Long sessions.
  • A capped keep-warm ping holds a large conversation's cache across idle gaps (TOKEN_OPTIMIZER_PROXY_KEEPWARM=0).
  • The Stop hook's harvest no longer takes a turn of its own (TOKEN_OPTIMIZER_HARVEST_TURN=1 restores it).

Written to Claude Code's settings

• autoCompactWindow: 400000, only when the key is absent. It is recorded, and removed again only while it still holds the value we wrote (TOKEN_OPTIMIZER_COMPACT_WINDOW=off, or mode off).
• The record of this and of default routing now sits beside the settings file it describes (token-optimizer-compact-window.json, token-optimizer-routing.json), not in the optimizer home. A record a 7.6.0 install left in ~/.token-optimizer is still read, and moved on the next write.

Changed for a Claude Code plugin install

• The MCP server no longer tells agents to load its tools with ToolSearch for ordinary reads and searches, because the hooks already compact that output. Other clients keep the routing policy.
• The post-call hook no longer runs on Read, Agent or Task (it runs on Grep, Bash, PowerShell and edits).

Built but off by default, because measuring showed they cost more or never fired:

• read-only model routing (TOKEN_OPTIMIZER_PROXY_READONLY_ROUTE=down);
• issue/PR prefetch (TOKEN_OPTIMIZER_PROXY_PREFETCH=1);
• server-side clearing and compaction (TOKEN_OPTIMIZER_PROXY_CLEAR_TOOL_USES=1, TOKEN_OPTIMIZER_PROXY_COMPACT=1);
• lean handoffs (TOKEN_OPTIMIZER_LEAN_HANDOFF=1, which also needs Agent|Task in the PostToolUse matcher);
• repo_ask (TOKEN_OPTIMIZER_REPO_ASK=1);
• the batching instruction (TOKEN_OPTIMIZER_BATCH_GUIDANCE=1).

Both of the on rows said off here until 7.1.0, which was wrong about the shipped code rather than a change of default: an unset value has always meant enabled, and 0, false, no and off are what turn either one off.

TOKEN_OPTIMIZER_PROXY_NET_SAVING is off because the knowledge block is supposed to grow the request: it is charged once as a cache write and then read at 0.1x for the rest of the session, which pays on a long conversation and does not on a short one. Turn it on and the block is dropped on any turn where sending it would hand upstream more bytes than the client handed us -- the wire never inflates, and the prefix knowledge is what you give up for that.

TOKEN_OPTIMIZER_PROXY_CLEAR_TOOL_USES is off because the client re-sends its whole history every turn: once a conversation passes the trigger, every request clears, and the boundary moves forward with each new tool use, re-writing the cache from there each time while the model re-reads what was cleared. On the long single-agent benchmark the run that crossed the trigger cost about 1.5x the runs without clearing. TOKEN_OPTIMIZER_REPO_ASK is off because agents offered it never used it (0 calls across about 230 benchmark subagents); the briefing line was a cost on every agent. TOKEN_OPTIMIZER_PROXY_PREFETCH is off because on the 10-agent campaign replay the workers fetched their issues themselves anyway, and runs with it cost about 10% more than runs without. TOKEN_OPTIMIZER_PROXY_READONLY_ROUTE is off because the routed subagents searched in many more, smaller calls and every call re-reads the whole context: on the exploration benchmark Claude Code reported $0.349 per run routed against $0.255 unrouted.

Rollout channels

Each switch above is its own decision, which is fine until there are ten of them. A channel is one decision instead: stable is what is on by default today, and each wider channel carries everything the narrower ones do plus the features still being measured.

variable	default	what it does
TOKEN_OPTIMIZER_ROLLOUT_CHANNEL	stable	stable, beta, canary, dev
TOKEN_OPTIMIZER_FEATURES	unset	a comma list of feature names to turn on within the channel
TOKEN_OPTIMIZER_DISABLE_FEATURES	unset	a comma list to turn off, whatever else is set

beta turns on install self-repair; canary adds compact tool definitions. Those two are staged: the channel is what enables them.

Everything else a wider channel reaches is request-only, and a channel never turns one on for you -- it only lets you ask. beta makes the output shaper and the shared knowledge graph askable, canary adds identifier substitution and thinking-block dropping, and dev holds nothing back. Each of those either changes what the model writes or forces an answer the code otherwise works out from the directory it is running in, which is a decision worth naming rather than inheriting:

TOKEN_OPTIMIZER_ROLLOUT_CHANNEL=canary TOKEN_OPTIMIZER_FEATURES=substitution

A variable you already set still wins over the channel. If TOKEN_OPTIMIZER_PROXY_SUBSTITUTE=1 is in your settings, substitution is on even on stable, because that switch shipped before channels existed and moving you to a channel is not supposed to quietly turn your own configuration off. TOKEN_OPTIMIZER_DISABLE_FEATURES is the one thing nothing overrides.

A name this does not recognise turns nothing on, and a value it cannot read is not treated as yes. Both are reported rather than guessed at:

npm run doctor           # the channel and what it turned on
npm run doctor -- --features   # every feature and the reason for each

The telemetry switches

variable	default	what it does
TOKEN_OPTIMIZER_TELEMETRY	off	aggregate counts into a file in your home directory
TOKEN_OPTIMIZER_BEACON	off	upload that file; requires the one above as well
DO_NOT_TRACK	unset	overrides both, whatever else is set
TOKEN_OPTIMIZER_BEACON_URL	project	send somewhere else, for a fork or a self-hosted table
TOKEN_OPTIMIZER_BEACON_KEY	packed	the key to send with; empty in a build without one
TOKEN_OPTIMIZER_BEACON_TABLE	telemetry_events	the table to insert into
TOKEN_OPTIMIZER_UPDATE_CHECK	on	ask npm for the latest version, in doctor only

The key is not in this repository. It is public and the package is published, so a committed key would be handed to everyone who runs npm view and could not be rotated out of the copies already installed; the release workflow stamps it into the tarball instead. A build without it still works and still aggregates locally -- it just cannot upload, and install_doctor prints packed without a key rather than letting an opt-in look like it is working.

lossless is worth knowing about: it forbids every transform that removes something the output cannot reconstruct -- function bodies, array tails, lower-signal prose -- and keeps the ones that can, which is whitespace, null keys, folded duplicate lines with their timestamps listed, repeated path prefixes, and back-references to content still in the request. Measured on the codebase-exploration workload it still removes 12.2%, with zero lossy elisions. For review or audit work where "the model can read the path" is not an acceptable answer, that is the setting.

Presets are a starting point, not a ceiling: the library takes a CompressionOptions object layered over the preset, so "aggressive, but keep six head rows" is expressible.

The graph, in the cached prefix

The expensive failure here is turns, not tokens. This project measured a posture that cut nothing and cost 1.471x through extra turns alone. So a finding that prevents one wasted turn pays for a great deal of context.

The knowledge graph knows things that would prevent them, but until now it reached the model two ways and both arrive too late or too dear: a SessionStart index chosen once from the opening task text, and a PreToolUse advisory that fires after the model already decided to make the call it is advising about -- so acting on it costs the very turn it was meant to save.

Verified project findings enter the cached prefix by default, with a 2,000-character budget. The proxy freezes the selected block for a conversation to preserve the cache; new sessions can select newly recorded findings. This includes the Codex Responses API path. Shared graphs exclude project-specific claims. Set TOKEN_OPTIMIZER_PROXY_KNOWLEDGE=0 to disable injection.

Added knowledge is reported separately as injectedChars. It adds input tokens; net cost improvement requires measuring whether it prevents enough work. Earlier head-to-head results do not establish that the newly enabled combination wins every workload.

Images

An image block was invisible to all of this until recently -- every walker keyed on block.text. That matters because an image costs about width * height / 750 tokens, so one 1456x816 screenshot is roughly 1,585, re-sent as history on every later turn. A browser-driving agent sends the same frame repeatedly, and those repeats now collapse to a reference to the copy already in the request. Dimensions come from the file header, so there is no codec dependency; resizing is deliberately not done for the same reason.

Relevance: BM25 by default, a model if you want one

Retention is ranked by BM25, which is why this needs no Python, no model weights and no RAM floor, and why it is deterministic enough to keep a cached prefix byte-stable. Lexical has a ceiling, though: "the database ran out of handles" shares no token with "connection pool exhausted", and worse, BM25 does not go quiet on that question -- it confidently picks the wrong line.

So there are two ways to do better. registerRanker installs any synchronous ranker. Or supply a SemanticEncoder and let the proxy warm an embedding cache: a request-level async pre-pass embeds the candidate units in one batch, and the engines then read vectors synchronously, exactly as before. An onnxruntime-node adapter is included, loaded dynamically so the runtime is never a dependency of this package.

Anything the pre-pass did not see falls back to BM25 rather than scoring zero -- a missing vector is our omission, not evidence about the line.

The adapter is verified against real inference, not mocked: npm run verify:onnx runs nine checks against a 2.3 KB ONNX graph with real weights, committed alongside the generator that builds it. It is a script rather than a jest test because onnxruntime checks instanceof Float32Array in native code and jest's per-suite VM realm has its own typed arrays; CI installs the optional runtime and runs it.

Not claimed: that an embedding model actually beats BM25 on these workloads. The mechanism is proved and the measurement is not done.

Compaction is consolidation, not loss

Everyone else checkpoints and restores what you had — which spends the scarcest budget in the session replaying context you already paid for.

Selection here is derived, not a category list: cost-to-rederive × irrecoverability × reuse-probability, with dead ends and decisions on a floor, because cheap-to-find is not the same as cheap-to-find-again. Restoration then adapts to the situation — mid-problem, cold resume, or in-flow — within a measured budget:

Where you were: does clock skew explain the 401s? ruled out: token signing, clock drift on the client untested: NTP skew on the server

That is resuming a thought. A summary describes one.

Progressive disclosure that knows what you asked

A large tool result becomes a preview chosen by the session's actual question, after parsing the output's shape (test report, diff, stack trace, log, JSON) — not the first 40 lines because they are first.

[selected against: "which shard fails?"]
--- failures ---
  FAILED  DBNetTests.BceOnRelu -- expected 0.0 got NaN
  FAILED  TftGradientFlow -- gradient did not reach the encoder
---- omitted: 1,760 lines of passing tests (expand 8bb6bd66) ----

Every cut is named, because a model reasoning over a silent truncation cannot know it is missing anything. expand serves from a content-addressed store — it never re-runs your test suite — and expanding both teaches the next preview and promotes what you needed into the graph, so the second expansion never happens.

Prompt-cache economics, measured from your own transcript

Provider caches are billable and provider-specific; a cache hit is not a free input token. For example, Anthropic publishes separate cache-read and cache-write multipliers, while OpenAI and Gemini expose their own cached-input usage and pricing rules. Token Optimizer reads native cache fields when the client supplies them and keeps reads, writes, uncached input, and output separate. It never applies one provider's cache multiplier to another client. The attribution view then does the part a hit rate cannot:

! CLAUDE.md:2 has an embedded timestamp, invalidating everything after it
    about 329,421 tokens re-written per session

Attributed to a line, priced by what sits behind it. Keep-warm is decided by expected value from your observed gaps, per TTL tier — and when neither tier pays, it says so.

Model routing decided by outcomes, not by task size

Everyone guesses from task shape and never checks. This reads which model ran each episode and what happened — retries, errors, turns — and prices both mistakes: what an overpowered model wastes, and what an underpowered one costs in retries. A tier that needs a retry in more than half its episodes is excluded at any price, because four cheap turns that fail are not cheap.

Waste detection that becomes a ratchet

A report is read once and forgotten. Here a detection produces a durable, measured, reversible fix — a skip rule, a composite touch — plus a ~50-token session-start briefing so the waste never starts. Detectors are a shipped floor plus patterns derived from your project's own history, each carrying what it has actually saved:

generated/schema.d.ts: read in 9/9 sessions, never the source of a finding
    3,400 tokens/session; cost equivalent not priced; apply: waste_audit action="apply"

Anything that touches your files is proposed as a diff and never applied.

One audit across every project

fleet_audit ranks your whole machine by measured cost, and does something a per-project scanner cannot: a fix proven in one project is offered to the others containing the same file contents, carrying the evidence from where it was measured. Matching is by content hash, never by filename.

It also runs the natural experiment nobody else can — enforcing clients versus directive ones — and reports it whichever way it falls, with the confound stated.

It learns from what already went wrong, offline

Every failed tool call you have already paid for is sitting in your agent's own transcripts. npx token-optimizer-learn reads them locally, groups the failures that keep repeating, and writes the ones worth remembering into the instructions file that agent already reads -- CLAUDE.md for Claude Code, AGENTS.md for Codex.

npx token-optimizer-learn                  # what it found, printed, nothing written
npx token-optimizer-learn --projects       # which projects have transcripts
npx token-optimizer-learn --since 14 --write

--agent <claude|codex|auto>, --project <path>, --since <days>, --max-sessions <n>, --json. Nothing is written without --write, and a block it writes is delimited, so running it again replaces its own advice instead of appending another copy -- nothing outside those markers is ever touched.

Three things it refuses to do, because each one produced a confidently wrong rule on a real corpus:

• It does not blame a chained command line. A transcript records the line and the exit code, not which of cd X && git grep a && ls b failed, so those are counted as unattributable and reported as a number rather than pinned on the first command.
• It does not turn your failing build into advice. A failing test or compile is the work going wrong, not a habit to avoid.
• It says what it could not read. Transcripts it could not parse and failures it could not categorise are printed as counts, because they bound how much of the result to believe.

It reads local files only -- no network, no upload -- and reduces your home directory to ~ before any recommendation text is written. TOKEN_OPTIMIZER_LEARN_PLUGINS takes a comma-separated list of module specifiers to add another agent's format; nothing is discovered or imported unless you name it.

---

Trust: we ship hooks that refuse your tool calls

That is a bigger ask than a normal dependency makes, so:

Verify the release. Published from CI with npm provenance — npm audit signatures ties the artifact to the workflow run and the commit, without trusting us. CHECKSUMS.sha256 ships alongside for offline checking.

If nothing seems to be happening, the lifecycle bundle is probably not installed. An MCP server process cannot modify the host that launched it, and npm 11 gates lifecycle scripts behind allow-scripts. Install the native plugin/hook bundle listed for your client below; adding only the MCP server gives the model tools but no pre-execution veto. For a legacy global Claude Code installation, recovery is one line:

npx token-optimizer-install     # wire Claude Code hooks
npx token-optimizer-doctor      # prove the Claude hooks work

Check that it works — not that files exist.

npx token-optimizer-doctor      # or npm run doctor, from a clone

It feeds a synthetic payload to the real hook binary and asserts a large read is refused and a small one is not. A checklist would have passed on the exact bug this project once shipped, where the plugin was connected, visible in /mcp, and saving nothing. Every failure names its own fix.

Every refusal carries its own off switch. Enforcement that hides its disable is coercive, and the person who needs it is mid-refusal, not reading a README:

auth.ts is 91 KB. Call smart_read instead.
(Not what you wanted? TOKEN_OPTIMIZER_MODE=assist stops refusals and keeps everything else on.)

Removal is exact. The installer records every file it wrote, with hashes. Uninstall removes only what still matches; anything you edited since is left in place and named. Your own hooks are never touched, and we never rewrite your settings.json — we merge into it.

npm run uninstall-hooks              # show the plan; changes nothing
npm run uninstall-hooks -- --apply   # carry it out

---

It separates direct savings from causal graph evidence

Every tool in this space reports "tokens saved" computed from its own assumptions. That number cannot be wrong, because nothing checks it.

For optimizer results, this records the materialized before-state and the text actually returned to the client, then subtracts any linked expansion responses. Native graph-substitution counterfactuals stay labeled as modeled and outside the main headline.

The graph's broader claim—whether delivered knowledge prevented later re-reading—is causal, so it runs a randomized holdout. Delivery is silently withheld on a slice of file touches, stratified by file, and the effect is the difference in downstream reads between arms. That effect is not added to the combined net until the experiment can support it, and the page will tell you plainly:

the graph is NOT yet paying for itself

A tool that can only ever report good news is not reporting. The same discipline runs throughout: an unmeasurable saving renders as unknown, never as zero, and never as $0.00 — because "cannot tell yet" printed as "saved nothing" is a silent false negative, and dollars get quoted to other people.

It runs everywhere, and says which tier

The tier is a protocol guarantee, not a preference:

• Lifecycle continuation: Claude Code, Codex, GitHub Copilot CLI, Gemini CLI, Qwen Code, and Cursor. Native routing/capture/delivery plus one active-model completion reflection.
• Native observation: Cline, OpenCode, Kilo, and Windsurf. Native routing/capture/delivery; the active model performs semantic writes.
• MCP + rules: Roo Code, Zed, Amp, Continue, Crush, and Droid. MCP-visible activity and explicit graph tools, with no claim over hidden built-in calls.

The exact surfaces still differ. For example, a protocol that can replace a large read before it reaches the model provides a stronger token guarantee than one that can only observe it. The generated registry, adapters, dashboard, and certification report all read the same capability source so those claims cannot drift independently.

Every config shape is confirmed against that client's published documentation, with the source URL recorded in its README.

Everything here is verified, and you can run it

npm run verify:all

Suite	Checks	What it proves
test	2,437	Enforcement, staleness, injection, consolidation, disclosure, cache, routing, trust — driving the real hooks over stdin
verify:clients	253	Every client config, lifecycle manifest, and enforcement surface matches its documented schema
verify:harvest	26	Request shape, response parsing, and that no secret from a tool result crosses the wire
verify:ui	22	Real headless Chromium: layout, label collisions, legibility
doctor	10	The installed hooks actually refuse, and the server actually answers

These are not decoration. They found six client configs that would have failed silently, an installer that destroyed user hooks, a Windows path bug that made enforcement blind to half its own refusals, and an uninstaller that printed a plan and deleted nothing.

OrcaRouter

OrcaRouter is an OpenAI-compatible AI gateway that routes many providers behind one endpoint. It is wired here as a first-class provider for the optimizer's own model calls, with two independent ways in:

Choice	Provider id	What it does
OrcaRouter - API	orcarouter	You paste an sk-orca-… key you already hold. Nothing opens.
OrcaRouter - Auth	orcarouter-oauth	OAuth 2.0 + PKCE: a browser asks for consent and issues you a key.

Both produce the same ordinary OrcaRouter API key, and everything downstream — the Bearer header, the endpoint, model discovery, 401 recovery — is identical whichever one you used. The two are not interchangeable in the UI on purpose: a single button that sometimes asks for a key and sometimes opens a browser makes logout, reauthentication and support harder.

Configure it

Open the dashboard (npm run dashboard) and use the OrcaRouter card, or set it from the environment:

export ORCAROUTER_API_KEY=sk-orca-…          # the API-key choice
export TOKEN_OPTIMIZER_ORCA_MODEL=orcarouter/auto   # which model the optimizer calls

Or connect with an account instead, which writes the issued key under $TOKEN_OPTIMIZER_HOME (default ~/.token-optimizer) with owner-only permissions:

token-optimizer-route orcarouter --connect   # opens a browser, then stores the key
token-optimizer-route orcarouter --status    # shows a redacted status and the origins in use
token-optimizer-route orcarouter --clear     # removes the stored key

A PKCE-issued key is durable, not a refresh token. OrcaRouter returns a long-lived key, not an access/refresh pair, so it is reused until you revoke it at https://www.orcarouter.ai/console/authorized-apps — there is no refresh to schedule and re-authorizing on every launch would hit the account's limit of ten keys per 24 hours. When the relay answers 401, the exact account and credential generation that made the rejected request is marked for reauthentication, and nothing attempts a refresh. Refreshable OAuth tokens rotate automatically; durable key grants such as OrcaRouter are reused until the provider revokes them.

Origins

Authentication and inference are different hosts, and neither is derived from the other:

Purpose	Default
Consent screen and code exchange	https://www.orcarouter.ai
Inference and model discovery	https://api.orcarouter.ai/v1

https://api.orcarouter.ai/v1/auth/keys is a 404 — the auth endpoints are not under the relay. Self-hosted installs can set ORCA_BASE_URL for a single shared origin, or ORCA_AUTH_BASE_URL / ORCA_API_BASE_URL separately; explicit values win, and cleartext HTTP is permitted only for loopback.

Models

The model control is a listbox built from GET /v1/models on the configured origin, using your own key so the list is what your workspace can actually call. Model ids keep their vendor/model namespace. Options are filtered per entry point, and a model that does not declare a capability is not offered rather than offered and then rejected:

• text chat — ?capability=chat, and the model must declare an openai, anthropic, gemini or openai-response endpoint type;
• multimodal — the same, plus an explicit architecture.input_modalities entry for the modality being sent (an undeclared capability fails closed);
• embedding, image, video and rerank — matched strictly against the embeddings, image-generation, openai-video and jina-rerank endpoint types.

If discovery fails, a small verified fallback list is shown and labelled as such (openai/gpt-5.5, anthropic/claude-opus-4.8, google/gemini-3.5-flash, deepseek/deepseek-v4-pro, orcarouter/auto), with its context, modality and reasoning-effort metadata intact. A successful live result is authoritative and never has the fallback mixed into it.

Text entry point	After switching to images

The second capture is after the entry point is switched to images: the text-only selection is cleared, and only models whose catalog entry declares image input remain.

npm run evidence:orcarouter regenerates the screenshots and assertions for the card into orca-evidence/ (untracked — it is a run artifact). The copies linked from this README live in docs/media/orcarouter/, alongside the other dashboard captures. A delivery check reads orca-evidence/ and refuses a patch that ships it, so the directory is ignored rather than committed.

---

Honest comparison

	Token Optimizer	Typical alternatives
License	MIT — commercial use fine	Often noncommercial-only; check before using at work
Default behaviour	Refuses the wasteful call	Suggests a better tool
Re-read of an unchanged file	Returns a diff	Returns the file again
Savings figure	Direct before/return ledger; causal graph effect separately held out	Computed from the tool's own assumptions
Cross-session memory	Findings, decisions, dead ends	Usually none
Compaction	Consolidation, ranked by cost-to-rederive	Checkpoint and replay
Cache economics	Measured from the transcript, attributed to a line	Rarely addressed
Model routing	Measured from episode outcomes	Guessed from task size
Cross-project	Fixes transfer by content hash	Per-project only
Clients	16	3–6 typical
Telemetry	None	Varies

---

Installation

Every client launches the same stdio server — npx -y @ooples/token-optimizer-mcp@latest — but they differ in what they let a hook do, and that difference is the whole product. A client with a pre-execution veto can have the wasteful call refused; one without can only be told. Both are listed honestly below.

Ready-made configuration for all sixteen lives in integrations/, generated from one source and validated by npm run verify:clients. Full matrix: docs/CLIENT_SUPPORT.md.

Every MCP connection also receives capability-aware mandatory routing instructions in its initialize response. That gives all clients a universal always-on policy, but only the ten clients with native pre-tool surfaces can hard-veto a wasteful built-in call; install their lifecycle bundle for actual enforcement.

Enforcing tier — the wasteful call is refused

These ten clients expose a pre-execution hook. Their packaged lifecycle bundle shares one capability-aware decision engine and defaults to assist: routing, retrieval, capture and harvest all on, no refusals. Set TOKEN_OPTIMIZER_MODE=enforce if you want expensive built-in calls vetoed, or off to disable the hooks entirely.

Client	Installable lifecycle surface
Claude Code	Native plugin: /plugin marketplace add ooples/token-optimizer-mcp, then /plugin install token-optimizer@token-optimizer
Codex	Native plugin in integrations/codex/plugin, or standalone hooks in integrations/codex
GitHub Copilot CLI	Project hooks in integrations/copilot
Gemini CLI	Gemini extension at the repository root, backed by integrations/gemini
Qwen Code	Extension bundle in integrations/qwen
Cursor	Project hooks and always-applied rule in integrations/cursor
Cline	Project hooks and rule in integrations/cline
OpenCode	In-process plugin and hooks in integrations/opencode
Kilo	In-process plugin and hooks in

Rules tier — mandatory routing where the host has no veto API

Roo Code, Zed, Amp, Continue, Crush, and Droid do not expose a packaged pre-execution bridge that can safely veto built-in calls. Their generated, always-on rules make optimized routing mandatory whenever the exact MCP schema is visible, and fail open to a bounded native operation when it is not. The integration directories contain both the MCP config and the rules file at the paths documented by each host.

These paths are checked, not assumed. Verifying them against each client's published docs found six configs that would have installed cleanly and never loaded — including Kilo, whose schema shares nothing with the mcpServers convention the other clients use. A convention is not a schema.

Codex

1. Add the MCP server or native plugin

For the best experience, install the Codex plugin. It bundles the MCP server, the token-optimization skill, session guidance, and a large-read hook:

codex plugin marketplace add ooples/token-optimizer-mcp
codex plugin add token-optimizer@token-optimizer

Review and trust the bundled hooks with /hooks, then start a new conversation. If you prefer an MCP-only installation, use:

codex mcp add token-optimizer -- npx -y @ooples/token-optimizer-mcp@latest

On Windows, if PowerShell blocks the codex.ps1 shim, use the command launcher directly:

codex.cmd mcp add token-optimizer -- npx -y @ooples/token-optimizer-mcp@latest

This writes the server to ~/.codex/config.toml. Codex CLI, the Codex IDE extension, and the Codex app on the same host share that configuration.

2. Verify the installation

codex mcp get token-optimizer
codex mcp list

Start a new Codex conversation after installation so the new tools are discovered. In an interactive CLI session, /mcp shows the tools available to the conversation.

3. Add optimization guidance and hooks

The plugin supplies both automatically. For an MCP-only installation, add the guidance from integrations/AGENTS.md to a project or global AGENTS.md. A ready-made standalone hook is also available under integrations/codex/hooks; merge its hooks.json into ~/.codex/hooks.json, copy the script to ~/.codex/hooks/, and review it once with /hooks.

The Codex hook injects guidance at SessionStart. Under TOKEN_OPTIMIZER_MODE=enforce it blocks expensive native operations when the bundled MCP has an exact replacement, including a single unambiguous code-mode shell call such as cat or Get-Content; multi-operation orchestration remains advisory so unrelated work is not discarded, and a second attempt at the same target passes through if the MCP is unavailable. The default is assist, which keeps retrieval and capture but never vetoes; TOKEN_OPTIMIZER_MODE=advise adds the routing advisory without vetoes, and off disables the hooks. The AGENTS.md/skill guidance remains important.

If you prefer a smaller instruction block:

## Token optimization

Use the token-optimizer MCP for large or repeated reads:

- `smart_read` for files over roughly 400 lines and for files already read once.
- `smart_glob`/`smart_grep` for large search results.
- `optimize_text` to store bulky text outside the model context.
- `get_optimization_report` when the user asks for token or compression stats.

Use normal tools for small, one-off operations.

See the current Codex hooks documentation for hook trust, matching, and tool-coverage details.

Equivalent manual Codex configuration

If you prefer to edit ~/.codex/config.toml yourself:

[mcp_servers.token-optimizer]
command = "npx"
args = ["-y", "@ooples/token-optimizer-mcp@latest"]

# Optional: keep the cache in a custom location.
# env = { TOKEN_OPTIMIZER_CACHE_DIR = "/absolute/path/to/cache" }

Claude Code

Install the plugin, not the bare MCP server. The plugin is the only path that optimizes by default; adding the MCP server alone gives the model a set of tools it is free to never call.

/plugin marketplace add ooples/token-optimizer-mcp
/plugin install token-optimizer@token-optimizer
/reload-plugins

That is the whole installation. There is nothing to configure and no flag to turn on.

What you get immediately

From the first message of the next session, expensive built-in calls are refused and redirected to the optimized equivalent:

You (or the model) do this	What happens
Read a file over ~25 KB	Denied → smart_read (cached)
Read any file already read this session	Denied → smart_read (returns only the diff)
Grep file contents / Glob for files	Denied → smart_grep / smart_glob
Edit a file over ~25 KB	Denied → smart_edit (returns a diff, not the file)
cat/head/tail/Get-Content a large file	Denied → smart_read
grep -r / rg across the tree	Denied → smart_grep
Context fills and compaction starts	optimize_session runs first

The re-read case is usually the largest single win and the one most often missed: a 5 KB config read fifteen times across a session costs far more than one 200 KB file read once. Size-based rules never catch it.

It cannot get you stuck

Three properties, all tested:

• Fail-open. Any error in the optimizer — bad payload, unreadable file, unexpected exception — allows the original call through, exactly as if the plugin were not installed.
• Loop-breaking. A given target is refused once. Come back to it and it is allowed. So if the MCP server is missing or misconfigured, the cost is one wasted turn per file, self-healing, with no intervention.
• Cheap calls are left alone. Small files, paged reads, git log | head, and binary paths are never touched.

Turning it down

One variable, no reinstall:

TOKEN_OPTIMIZER_MODE=advise   # nudge instead of refuse (the pre-5.2 behaviour)
TOKEN_OPTIMIZER_MODE=off      # disable the hooks entirely
TOKEN_OPTIMIZER_LARGE_READ_BYTES=51200   # raise the "large file" threshold

MCP server only (not recommended)

If you want the tools without the enforcement:

claude mcp add --transport stdio --scope user token-optimizer -- \
  npx -y @ooples/token-optimizer-mcp@latest

Verify with claude mcp get token-optimizer, or /mcp inside Claude Code. Then add the recommendations from integrations/AGENTS.md to your CLAUDE.md — but be aware that guidance in a context file is advisory, and models routinely read past it.

Installing the package changes nothing on your machine. Hooks and managed Claude Code, Codex, and OpenCode commands are wired by one explicit command:

npm install -g @ooples/token-optimizer-mcp@latest
token-optimizer-install

token-optimizer-install is the only thing that edits ~/.claude/settings.json; the install itself prints the path it would write and exits. The hooks it adds match nearly every tool call in every project on the machine, so that is not a side effect an npm install should have. Undo it with token-optimizer-uninstall, which is a dry run until you pass --apply. To activate as part of a container image or provisioning script, set TOKEN_OPTIMIZER_AUTO_INSTALL=1 before installing. Open a new shell to activate managed CLI commands, or use token-optimizer-run directly. See the hook installation guide.

GitHub Copilot CLI

1. Add the MCP server

On current Copilot CLI releases:

copilot mcp add token-optimizer -- npx -y @ooples/token-optimizer-mcp@latest

If your Copilot CLI does not expose copilot mcp yet, save this as ~/.copilot/mcp-config.json:

{
  "mcpServers": {
    "token-optimizer": {
      "type": "local",
      "command": "npx",
      "args": ["-y", "@ooples/token-optimizer-mcp@latest"],
      "tools": ["*"]
    }
  }
}

A ready-made copy is available at integrations/copilot/mcp-config.json.

2. Verify the installation

copilot mcp get token-optimizer
copilot mcp list

Inside an interactive Copilot session, /mcp show token-optimizer displays the connection status and available tools.

3. Add optimization guidance and hooks

Keep integrations/AGENTS.md as the repository's AGENTS.md, or adapt the same guidance into .github/copilot-instructions.md.

For native lifecycle integration, install the hooks with the bundled command. It copies the entry points AND composes the shared core beside them, which a plain cp -r cannot do: the package ships the core once, not ten vendored copies.

npx -y @ooples/token-optimizer-mcp@latest token-optimizer-install-client --client copilot

It writes .github/hooks/ by default, and --check reports what it would write without writing it. --dest <dir> writes to another directory but changes no client configuration, so it has to be the directory the client's own config already reads -- which is why the clients that resolve that directory at runtime (Gemini, Qwen, the Codex plugin) require it.

The hooks inject optimization guidance at sessionStart. Under TOKEN_OPTIMIZER_MODE=enforce they deny a large built-in view so Copilot retries with smart_read; the default assist leaves the call alone. Partial reads and files below 25 KB always pass through unchanged, and TOKEN_OPTIMIZER_MODE=advise gives recommendations without vetoes. Repository hooks work without overwriting user-level files; global hooks can instead be placed in ~/.copilot/hooks/ with their script paths adjusted for that directory.

Restart Copilot CLI after changing hook files. See GitHub's official MCP setup guide and hooks reference.

Gemini CLI

1. Add the MCP server

Add Token Optimizer directly at user scope:

gemini mcp add --scope user token-optimizer npx -y @ooples/token-optimizer-mcp@latest

Alternatively, install this repository as a Gemini extension so the MCP configuration and GEMINI.md guidance are packaged together:

gemini extensions install https://github.com/ooples/token-optimizer-mcp --auto-update

2. Verify the installation

gemini mcp list
gemini extensions list

Run /mcp inside Gemini CLI to inspect the connection. Restart Gemini CLI after installing or updating the extension.

3. Add optimization guidance and hooks

Direct MCP users should copy GEMINI.md into the project or merge its rules into an existing GEMINI.md. Extension users receive that context file plus native hooks automatically.

The extension's SessionStart hook injects optimization guidance. Its AfterTool hook notices full-file read_file results over 25 KB and suggests smart_read. To make Gemini automatically replace those large results with a token-optimizer tail call, configure the extension setting Automatic large-read routing as true:

gemini extensions config token-optimizer

Automatic routing uses Gemini's native tailToolCallRequest: the smart_read result replaces the built-in read result before it reaches the model. Partial reads remain unchanged. Restart Gemini CLI after installing, updating, or reconfiguring the extension. See the official Gemini MCP guide, extension guide, and hooks reference.

OpenCode

1. Add the MCP server

Create or update opencode.json in your project:

{
  "$schema": "https://opencode.ai/config.json",
  "mcp": {
    "token-optimizer": {
      "type": "local",
      "command": ["npx", "-y", "@ooples/token-optimizer-mcp@latest"],
      "enabled": true
    }
  },
  "instructions": ["./AGENTS.md"]
}

For a global installation, merge the same mcp entry into ~/.config/opencode/opencode.json.

2. Verify the installation

opencode mcp list

The output shows configured servers and their connection status.

3. Add optimization guidance and the local plugin

Copy integrations/AGENTS.md to the project as AGENTS.md; the instructions entry above loads it. Then copy the ready-made local plugin:

mkdir -p .opencode/plugins
cp integrations/opencode/.opencode/plugins/token-optimizer.js .opencode/plugins/

New-Item -ItemType Directory -Force .opencode/plugins | Out-Null
Copy-Item integrations/opencode/.opencode/plugins/token-optimizer.js .opencode/plugins/

On OpenCode 1.3.3 or earlier (check opencode --version), copy integrations/opencode/legacy/.opencode/plugins/token-optimizer.js to the same place instead. Those releases call every export of a plugin file as a plugin function, so they cannot load the current file, whose default export is the object OpenCode 2.x requires. Both files have the same name, so copying one replaces the other.

The plugin preserves Token Optimizer usage state in OpenCode's compaction prompt. Under TOKEN_OPTIMIZER_MODE=enforce its tool.execute.before hook rejects full-file reads over 25 KB and steers the agent to smart_read; the default assist lets them through. Small and partial reads always pass normally, and TOKEN_OPTIMIZER_MODE=advise gives non-blocking guidance. Restart OpenCode after adding the plugin. See the official OpenCode MCP guide and plugin hook guide.

Generic MCP configuration

Any stdio-capable MCP client can launch Token Optimizer with:

{
  "mcpServers": {
    "token-optimizer": {
      "command": "npx",
      "args": ["-y", "@ooples/token-optimizer-mcp@latest"]
    }
  }
}

Additional ready-made integration files are available for Claude Desktop, Codex, Gemini CLI, OpenCode, and GitHub Copilot.

Use it

You normally use Token Optimizer by asking your agent in plain language:

Use token-optimizer smart_read for the large server file, then use it again
after the edit so only the diff comes back.

Cache this API response with optimize_text under the key customer-schema,
then retrieve it only if we need the full payload again.

Show my token savings with get_optimization_report.

For clients that expose direct MCP tool calls, the core inputs are small JSON objects:

{
  "tool": "smart_read",
  "arguments": {
    "path": "/absolute/path/to/large-file.ts"
  }
}

{
  "tool": "optimize_text",
  "arguments": {
    "text": "A large response, log, document, or generated artifact...",
    "key": "stable-reference-key",
    "quality": 11
  }
}

{
  "tool": "get_optimization_report",
  "arguments": {
    "topN": 10
  }
}

Understand the compression stats

optimize_text returns measurements with every call. This example uses a deliberately repetitive payload to make every field easy to see; it is not a benchmark:

{
  "success": true,
  "key": "customer-schema",
  "originalTokens": 4180,
  "compressedTokens": 72,
  "tokensSaved": 4108,
  "percentSaved": 99.55,
  "cached": true,
  "compressionUsed": true
}

get_optimization_report applies the same versioned measurement contract as the dashboard and aggregates qualifying operations into:

• gross preview reduction, linked expansion debits, and net transport avoided;
• observed returned context and the net reduction percentage;
• verified operations tracked, with legacy and tool-reported claims excluded;
• breakdowns by action/tool, hook phase, and MCP server;
• optional date-range and session filters.

Two tools sound similar but serve different purposes:

Tool	Use it for	Context-window effect
optimize_text	Store bulky text under a key and return a compact reference	Reduces text kept in the active context
compress_text	Produce Brotli/base64 data for storage or transport	May use more model tokens if pasted back into context

If your goal is a smaller prompt, prefer optimize_text. Use compress_text only when you specifically need byte compression.

What is included

Capability	Representative tools	What gets smaller or faster
Context and compression	optimize_text, get_cached, count_tokens, analyze_optimization, context_delta	Large payloads and repeated context
File and Git operations	smart_read, smart_write, smart_edit, smart_grep, smart_glob, smart_diff, smart_status	File contents, search results, and diffs
Caching	smart_cache, cache_warmup, cache_invalidation, cache_compression, predictive_cache	Repeated computation and retrieval
APIs and databases	smart_api_fetch, smart_sql, smart_graphql, smart_rest, smart_schema	Responses, schemas, and query analysis
Build and system tasks	smart_build, smart_test, smart_lint, smart_logs, smart_processes	Build logs and diagnostic output
Intelligence	smart-summarization, pattern-recognition, natural-language-query, recommendation-engine	Analysis and summaries
Analytics	get_optimization_report, get_action_analytics, get_hook_analytics, export_analytics	Token-savings visibility

See docs/TOOLS.md for detailed tool inputs and examples.

Requirements and data

• Node.js 22 or newer
• npm 9 or newer
• An MCP client with stdio transport support

Default local data locations include:

• cache: ~/.token-optimizer-cache/
• analytics: ~/.token-optimizer-mcp/analytics.db
• sessions and configuration: ~/.token-optimizer/

Set TOKEN_OPTIMIZER_CACHE_DIR to override the cache location.

Verify it, check it, remove it

This installs hooks that refuse your tool calls. That is a bigger ask than a normal dependency makes, so here is everything needed to check it and undo it.

Verify the release is genuine. The package is published from CI with npm provenance, which signs an attestation binding the artifact to the workflow run and the commit that built it:

npm audit signatures

That verifies without trusting us. A CHECKSUMS.sha256 is attached to each GitHub release for offline checking (sha256sum -c CHECKSUMS.sha256) — useful for mirrors, but weaker: it shares a trust root with the thing it hashes.

Check it actually works. Not that the files are in place — that it works:

npm run doctor

This feeds a synthetic payload to the real hook binary and asserts a large read is refused and a small one is not, that session-start emits the policy, that the graph directory is writable, and that the MCP server starts and lists its tools. Every failure names its own fix. There is also an install_doctor MCP tool.

Turn enforcement off, instantly. Every refusal says this, so you never have to come back here to find it:

TOKEN_OPTIMIZER_MODE=off      # no enforcement, no hooks
TOKEN_OPTIMIZER_MODE=advise   # suggestions only, nothing is ever denied

Remove it. The installer records every file it wrote, with hashes, so removal is exact rather than best-effort:

npm run uninstall-hooks              # show the plan; changes nothing
npm run uninstall-hooks -- --apply   # carry it out

It removes only files that still match what we wrote. Anything you have edited since is left in place and named, because removing it would destroy your work and removing it silently would be worse. Hooks you added yourself are not in the manifest and are never touched. Config entries we added are listed for you to remove — we do not rewrite your settings.json.

Technical reference

The detailed operational material below is intentionally retained for users who want to understand the complete tool surface, hooks pipeline, performance controls, analytics, and troubleshooting behavior.

Complete Tool Reference (74 Total)

Core Caching & Optimization (8 tools)

• optimize_text - Compress and cache text (primary tool for token reduction)
• get_cached - Retrieve previously cached text
• compress_text - Compress text using Brotli
• decompress_text - Decompress Brotli-compressed text
• count_tokens - Count tokens using tiktoken (GPT-4 tokenizer)
• analyze_optimization - Analyze text and get optimization recommendations
• get_cache_stats - View cache hit rates and compression ratios
• clear_cache - Clear all cached data

Usage Example:

// Cache large content to remove it from context window
optimize_text({
  text: 'Large API response or file content...',
  key: 'api-response-key',
  quality: 11,
});
// Result: 60-90% token reduction

Smart File Operations (10 tools)

Optimized replacements for standard file tools with intelligent caching and diff-based updates:

• smart_read - Read files with 80% token reduction through caching and diffs
• smart_write - Write files with verification and change tracking
• smart_edit - Line-based file editing with diff-only output (90% reduction)
• smart_grep - Search file contents with match-only output (80% reduction)
• smart_glob - File pattern matching with path-only results (75% reduction)
• smart_diff - Git diffs with diff-only output (85% reduction)
• smart_branch - Git branch listing with structured JSON (60% reduction)
• smart_log - Git commit history with smart filtering (75% reduction)
• smart_merge - Git merge management with conflict analysis (80% reduction)
• smart_status - Git status with status-only output (70% reduction)

Usage Example:

// Read a file with automatic caching
smart_read({ path: '/path/to/file.ts' });
// First read: full content
// Subsequent reads: only diff (80% reduction)

API & Database Operations (10 tools)

Intelligent caching and optimization for external data sources:

• smart_api_fetch - HTTP requests with caching and retry logic (83% reduction on cache hits)
• smart-cache-api - API response caching with TTL/ETag/event-based strategies
• smart_database - Database queries with connection pooling and caching (83% reduction)
• smart_sql - SQL query analysis with optimization suggestions (83% reduction)
• smart_schema - Database schema analysis with intelligent caching
• smart_graphql - GraphQL query optimization with complexity analysis (83% reduction)
• smart_rest - REST API analysis with endpoint discovery (83% reduction)
• smart_orm - ORM query optimization with N+1 detection (83% reduction)
• smart_migration - Database migration tracking (83% reduction)
• smart_websocket - WebSocket connection management with message tracking

Usage Example:

// Fetch API with automatic caching
smart_api_fetch({
  method: 'GET',
  url: 'https://api.example.com/data',
  ttl: 300,
});
// Cached responses: 95% token reduction

Build & Test Operations (10 tools)

Development workflow optimization with intelligent caching:

• smart_build - TypeScript builds with diff-based change detection
• smart_test - Test execution with incremental test selection
• smart_lint - ESLint with incremental analysis and auto-fix
• smart_typecheck - TypeScript type checking with caching
• smart_install - Package installation with dependency analysis
• smart_docker - Docker operations with layer analysis
• smart_logs - Log aggregation with pattern filtering
• smart_network - Network diagnostics with anomaly detection
• smart_processes - Process monitoring with resource tracking
• smart_system_metrics - System resource monitoring with performance recommendations

Usage Example:

// Run tests with caching
smart_test({
  onlyChanged: true, // Only test changed files
  coverage: true,
});

Advanced Caching (10 tools)

Enterprise-grade caching strategies with 87-92% token reduction:

• smart_cache - Multi-tier cache (L1/L2/L3) with 6 eviction strategies (90% reduction)
• cache_warmup - Intelligent cache pre-warming with schedule support (87% reduction)
• cache_analytics - Real-time dashboards and trend analysis (88% reduction)
• cache-benchmark - Performance testing and strategy comparison (89% reduction)
• cache_compression - 6 compression algorithms with adaptive selection (89% reduction)
• cache_invalidation - Dependency tracking and pattern-based invalidation (88% reduction)
• cache_optimizer - ML-based recommendations and bottleneck detection (89% reduction)
• cache_partition - Sharding and consistent hashing (87% reduction)
• cache_replication - Distributed replication with conflict resolution (88% reduction)
• predictive_cache - ML-based predictive caching with ARIMA/LSTM (91% reduction)

Usage Example:

// Configure multi-tier cache
smart_cache({
  operation: 'configure',
  evictionStrategy: 'LRU',
  l1MaxSize: 1000,
  l2MaxSize: 10000,
});

Monitoring & Dashboards (7 tools)

Comprehensive monitoring with 88-92% token reduction through intelligent caching:

• alert_manager - Multi-channel alerting (email, Slack, webhook) with routing (89% reduction)
• metric_collector - Time-series metrics with multi-source support (88% reduction)
• monitoring_integration - External platform integration (Prometheus, Grafana, Datadog) (87% reduction)
• custom_widget - Dashboard widgets with template caching (88% reduction)
• data_visualizer - Interactive visualizations with SVG optimization (92% reduction)
• health_monitor - System health checks with state compression (91% reduction)
• log_dashboard - Log analysis with pattern detection (90% reduction)

Usage Example:

// Create an alert
alert_manager({
  operation: 'create-alert',
  alertName: 'high-cpu-usage',
  channels: ['slack', 'email'],
  threshold: { type: 'above', value: 80 },
});

System Operations (6 tools)

System-level operations with smart caching:

• smart_cron - Scheduled task management (cron/Windows Task Scheduler) (85% reduction)
• smart_user - User and permission management across platforms (86% reduction)
• smart_ast_grep - Structural code search with AST indexing (83% reduction)
• get_session_stats - Session-level token usage statistics
• analyze_project_tokens - Project-wide token analysis and cost estimation
• optimize_session - Compress large file operations from current session

Usage Example:

// View session token usage
get_session_stats({});
// Result: Detailed breakdown of token usage by tool

Intelligence & Summarization (6 tools)

• intelligent-assistant - Context-aware assistance with compact recommendations
• natural-language-query - Natural-language querying over structured data
• pattern-recognition - Pattern discovery with summarized findings
• predictive-analytics - Predictive analysis with concise results
• recommendation-engine - Ranked, context-aware recommendations
• smart-summarization - Token-aware summarization for large content

Token Analytics (5 tools)

• get_optimization_report - Complete savings report with totals and breakdowns
• get_action_analytics - Savings aggregated by action or tool
• get_hook_analytics - Savings aggregated by hook phase
• get_mcp_server_analytics - Savings aggregated by MCP server
• export_analytics - Export recorded analytics for external analysis

Context State & Storage (2 tools)

• optimization_storage - Persist and retrieve optimized content
• context_delta - Track compact context changes between states

Architecture and Global Hooks

Native client lifecycle coverage

The MCP server is identical in every client, but lifecycle APIs are not. The repository ships client-native adapters instead of copying Claude event names into tools that would silently ignore them.

Client	Native integration events	Refusal under MODE=enforce	Advisory escape hatch
Codex	SessionStart, PreToolUse	Deny replaceable large reads and single shell dumps	TOKEN_OPTIMIZER_MODE=advise
Claude Code	PreToolUse plus optional global pipeline	Deny replaceable large reads and noisy searches	TOKEN_OPTIMIZER_MODE=advise
GitHub Copilot CLI	sessionStart, preToolUse, postToolUse	Deny large view calls and steer to smart_read	TOKEN_OPTIMIZER_MODE=advise
Gemini CLI	SessionStart, BeforeTool, AfterTool	Deny replaceable large reads before they enter context	TOKEN_OPTIMIZER_MODE=advise
OpenCode	tool.execute.before, compaction hook	Reject large full-file reads and steer to smart_read	TOKEN_OPTIMIZER_MODE=advise

The default is assist: routing, retrieval, capture and harvest are on and nothing is ever refused. That is the posture two independent harnesses measured as our best -- on THOL, assist scored 0.971 in 14.4 turns against control's 0.969 in 16.2, while enforce scored 0.960 in 20.3 turns and cost 1.471x control per task (median across 17 tasks; cheaper on only 2 of them). Set TOKEN_OPTIMIZER_MODE=enforce for the refusals described above, which use a 25,600-byte threshold; override it with TOKEN_OPTIMIZER_LARGE_READ_BYTES, use TOKEN_OPTIMIZER_MODE=advise for the routing advisory without vetoes, or TOKEN_OPTIMIZER_MODE=off to disable the lifecycle integration. Partial reads always pass through because they may already be more efficient than a full cached read, and under enforce one repeated attempt is allowed so a failed MCP server cannot permanently block work.

Analytics workflow and storage

Granular token usage analytics for pinpointing optimization opportunities:

• get_hook_analytics - Token usage breakdown by hook phase (PreToolUse, PostToolUse, etc.)
• get_action_analytics - Token usage breakdown by tool/action (Read, Write, Grep, etc.)
• get_mcp_server_analytics - Token usage breakdown by MCP server (token-optimizer, filesystem, etc.)
• export_analytics - Export analytics data in JSON or CSV format with filtering

Usage Example:

// Get per-hook analytics
get_hook_analytics({
  startDate: '2025-01-01T00:00:00Z',
  endDate: '2025-12-31T23:59:59Z',
});
// Result: Shows which hooks consume the most tokens

// Get per-action analytics
get_action_analytics({});
// Result: Shows which tools use the most tokens

// Export analytics as CSV
export_analytics({
  format: 'csv',
  hookPhase: 'PreToolUse',
});
// Result: CSV export filtered by PreToolUse hook

Key Features:

• Per-hook phase tracking (PreToolUse, PostToolUse, SessionStart, etc.)
• Per-action tracking (Read, Write, count_tokens, etc.)
• Per-MCP-server tracking (token-optimizer, filesystem, GitHub, etc.)
• Date range filtering
• JSON and CSV export
• Persistent storage with SQLite
• Zero performance impact (async batched writes)

Global Hooks System (7-Phase Optimization)

This complete seven-phase pipeline applies to the optional Claude Code global-hook installation. Codex, Copilot, Gemini, and OpenCode use the smaller native adapters above because their event names, payloads, and result-replacement capabilities differ.

When global hooks are installed, token-optimizer-mcp runs automatically on every tool call:

┌─────────────────────────────────────────────────────────────┐
│ Phase 1: PreToolUse - Tool Replacement                      │
│ ├─ Read   → smart_read   (80% token reduction)             │
│ ├─ Grep   → smart_grep   (80% token reduction)             │
│ └─ Glob   → smart_glob   (75% token reduction)             │
└─────────────────────────────────────────────────────────────┘
                          ↓
┌─────────────────────────────────────────────────────────────┐
│ Phase 2: Input Validation - Cache Lookups                   │
│ └─ get_cached checks if operation was already done          │
└─────────────────────────────────────────────────────────────┘
                          ↓
┌─────────────────────────────────────────────────────────────┐
│ Phase 3: PostToolUse - Output Optimization                  │
│ ├─ optimize_text for large outputs                          │
│ └─ compress_text for repeated content                       │
└─────────────────────────────────────────────────────────────┘
                          ↓
┌─────────────────────────────────────────────────────────────┐
│ Phase 4: Session Tracking                                   │
│ └─ Log all operations to operations-{sessionId}.csv         │
└─────────────────────────────────────────────────────────────┘
                          ↓
┌─────────────────────────────────────────────────────────────┐
│ Phase 5: UserPromptSubmit - Prompt Optimization             │
│ └─ Optimize user prompts before sending to API              │
└─────────────────────────────────────────────────────────────┘
                          ↓
┌─────────────────────────────────────────────────────────────┐
│ Phase 6: PreCompact - Pre-Compaction Optimization           │
│ └─ Optimize before Claude Code compacts the conversation    │
└─────────────────────────────────────────────────────────────┘
                          ↓
┌─────────────────────────────────────────────────────────────┐
│ Phase 7: Metrics & Reporting                                │
│ └─ Track token reduction metrics and generate reports       │
└─────────────────────────────────────────────────────────────┘

Performance evidence

Reduction targets in tool descriptions are design goals, not production measurements. They never enter the verified ledger. Use the dashboard or get_optimization_report for request-level measurements, and npm run dashboard:audit-savings to inspect both qualifying and excluded rows. No universal dollar value is claimed: configure an effective rate only when it reflects your own provider, model, cache, route, plan, tier, and credits.

Usage Examples

Basic Caching

// Cache large content to remove from context window
const result = await optimize_text({
  text: 'Large API response or file content...',
  key: 'cache-key',
  quality: 11,
});
// Result: Original tokens removed, only cache key remains (~50 tokens)

// Retrieve later
const cached = await get_cached({ key: 'cache-key' });
// Result: Full original content restored

Smart File Reading

// First read: full content
await smart_read({ path: '/src/app.ts' });

// Subsequent reads: only changes (80% reduction)
await smart_read({ path: '/src/app.ts' });

API Caching

// First request: fetch and cache
await smart_api_fetch({
  method: 'GET',
  url: 'https://api.example.com/data',
  ttl: 300,
});

// Subsequent requests: cached (95% reduction)
await smart_api_fetch({
  method: 'GET',
  url: 'https://api.example.com/data',
});

Session Analysis

// View token usage for current session
await get_session_stats({});
// Result: Breakdown by tool, operation, and savings

// Analyze entire project
await analyze_project_tokens({
  projectPath: '/path/to/project',
});
// Result: Cost estimation and optimization opportunities

Technology Stack

• Runtime: Node.js 22+
• Language: TypeScript
• Database: SQLite (better-sqlite3)
• Token Counting: tiktoken (GPT-4 tokenizer)
• Compression: Brotli (built-in Node.js)
• Caching: Multi-tier LRU/LFU/FIFO caching
• Protocol: MCP SDK (@modelcontextprotocol/sdk)

Supported AI Tools

Token Optimizer works with stdio-capable MCP clients and includes first-party setup guidance for:

• OpenAI Codex - Direct MCP setup or the bundled plugin, skill, and hooks
• Claude Code - Direct MCP setup or the bundled plugin, skill, and hooks
• GitHub Copilot CLI - MCP configuration, AGENTS.md guidance, and repository hooks
• Google Gemini CLI - Direct MCP setup or the bundled context-and-hooks extension
• OpenCode - Local MCP configuration, AGENTS.md instructions, and compaction plugin
• Claude Desktop, Cursor, Cline, and Windsurf - Standard MCP JSON configuration

See Installation for the supported commands and configuration files.

Performance Characteristics

• Compression Ratio: 2-4x typical (up to 82x for repetitive content)
• Context Window Savings: 60-90% average across all operations
• Cache Hit Rate: >80% in typical usage
• Operation Overhead: <10ms for cache operations (optimized from 50-70ms)
• Compression Speed: ~1ms per KB of text
• Hook Overhead: <10ms per operation (7x improvement from in-memory optimizations)

Performance Optimizations

The PowerShell hooks have been optimized to reduce overhead from 50-70ms to <10ms through:

• In-Memory Session State: Session data kept in memory instead of disk I/O on every operation
• Batched Log Writes: Operation logs buffered and flushed every 5 seconds or 100 operations
• Lazy Persistence: Disk writes only occur when necessary (session end, optimization, reports)

Environment Variables

Control hook behavior with these environment variables:

The MCP server exposes a 19-tool core catalog by default so tool schemas do not consume a large share of the model context. Set TOKEN_OPTIMIZER_TOOL_PROFILE=full before starting the server to expose all 104 specialized tools. TOKEN_OPTIMIZER_TOOL_PROFILE=core is the explicit form of the default. Live graph capture uses TOKEN_OPTIMIZER_TOOL_PROFILE=continuity, which exposes only capture and query. The extended cognitive profile adds checkpoints, outcomes, and receipt attestation. The native-token audit measures 480 startup tokens for continuity, 694 for extended cognitive, 4,815 for core, and 30,125 for full. Stateful consumers normally receive zero MCP tools: host pre-action delivery adds only the selected capsule through the client lifecycle channel. Other enabled MCP servers add their own schemas independently.

For file-focused work, opt into TOKEN_OPTIMIZER_TOOL_PROFILE=files to advertise seven tools: smart_read, smart_write, smart_edit, smart_glob, smart_grep, get_cached, and expand. This profile retains file and cache retrieval while omitting wiki, audit, and session tools. It works alongside the compression proxy; see the live comparison runner for the full-files configuration and measured tradeoffs.

The current hardened cross-CLI smoke does not qualify. In the final Codex-to-Claude adversarial pair, both successors were correct and the runtime capsule was delivered with exact provider model attestations and zero consumer MCP tools, but runtime used 5.383% more tokens and 41.758% more latency. The reciprocal runtime arm was blocked by Claude provider quota, and Antigravity 1.1.11 required account authentication before a Gemini model could run. These signed outcomes keep the release verdict insufficient; they are evidence of working provenance and fail-closed behavior, not powered effectiveness or universal superiority.

The replacement-grade protocol is intentionally larger than that smoke: 54,054 all-family/all-arm trial envelopes, 113,022 provider calls, three independent model families, direction-level non-inferiority, and 1,056 hard-negative opportunities per direction and arm so the Bonferroni-adjusted 95% family-wise false-delivery upper bound—not only the point estimate—must remain below 1%. Run npm run verify:ucr:study-design to validate the frozen metric coverage. The full protocol and CLI driver contract are in evals/ucr/FULL_STUDY_CONTRACT.md.

Performance Controls

• TOKEN_OPTIMIZER_USE_FILE_SESSION (default: false)

  • Set to true to revert to file-based session tracking (legacy mode)
  • Use if you encounter issues with in-memory session state
  • Example: $env:TOKEN_OPTIMIZER_USE_FILE_SESSION = "true"

• TOKEN_OPTIMIZER_SYNC_LOG_WRITES (default: false)

  • Set to true to disable batched log writes
  • Forces immediate writes to disk (slower but more resilient)
  • Use for debugging or if logs are being lost
  • Example: $env:TOKEN_OPTIMIZER_SYNC_LOG_WRITES = "true"

• TOKEN_OPTIMIZER_DEBUG_LOGGING (default: true)

  • Set to false to disable DEBUG-level logging
  • Reduces log file size and improves performance
  • INFO/WARN/ERROR logs still written
  • Example: $env:TOKEN_OPTIMIZER_DEBUG_LOGGING = "false"

Development Path

• TOKEN_OPTIMIZER_DEV_PATH
  • Path to local development installation
  • Automatically set to ~/source/repos/token-optimizer-mcp if not specified
  • Override for custom development paths
  • Example: $env:TOKEN_OPTIMIZER_DEV_PATH = "C:\dev\token-optimizer-mcp"

Performance Impact: Using in-memory mode (default) provides a 7x improvement in hook overhead:

• Before: 50-70ms per hook operation
• After: <10ms per hook operation
• 85% reduction in hook latency

Monitoring Token Savings

Real-Time Session Monitoring

To view your actual token SAVINGS, use the get_session_stats tool:

// View current session statistics with token savings breakdown
await get_session_stats({});

Output includes:

• Total tokens saved (this is the actual savings amount!)
• Token reduction percentage (e.g., "60% reduction")
• Cache hit rate and compression ratios
• Breakdown by tool (Read, Grep, Glob, etc.)
• Top 10 most optimized operations with before/after comparison

Example Output:

{
  "sessionId": "abc-123",
  "totalTokensSaved": 125430, // ← THIS is your savings!
  "tokenReductionPercent": 68.2,
  "originalTokens": 184000,
  "optimizedTokens": 58570,
  "cacheHitRate": 72.0,
  "byTool": {
    "smart_read": { "saved": 45000, "percent": 80 },
    "smart_grep": { "saved": 32000, "percent": 75 }
  }
}

Session Tracking Files

All operations are automatically tracked in session data files:

Location: ~/.claude-global/hooks/data/current-session.txt

Format:

{
  "sessionId": "abc-123",
  "sessionStart": "20251031-082211",
  "totalOperations": 1250, // ← Number of operations
  "totalTokens": 184000, // ← Cumulative token COUNT
  "lastOptimized": 1698765432,
  "savings": {
    // ← Auto-updated every 10 operations (Issue #113)
    "totalTokensSaved": 125430, // Tokens saved by compression
    "tokenReductionPercent": 68.2, // Percentage of tokens saved
    "originalTokens": 184000, // Original token count before optimization
    "optimizedTokens": 58570, // Token count after optimization
    "cacheHitRate": 42.5, // Cache hit rate percentage
    "compressionRatio": 0.32, // Compression efficiency (lower is better)
    "lastUpdated": "20251031-092500" // Last savings update timestamp
  }
}

New in v1.x: The savings object is now automatically updated every 10 operations, eliminating the need to manually call get_session_stats() for real-time monitoring. This provides instant visibility into token optimization performance.

How it works:

• Every 10 operations, the PowerShell hooks automatically call get_cache_stats() MCP tool
• Savings metrics are calculated from cache performance data (compression ratio, original vs compressed sizes)
• The session file is atomically updated with the latest savings data
• If the MCP call fails, the update is skipped gracefully without blocking operations

Note: For detailed per-operation analysis, use get_session_stats(). The session file provides high-level aggregate metrics.

Project-Wide Analysis

Analyze token usage across your entire project:

// Analyze project token costs
await analyze_project_tokens({
  projectPath: '/path/to/project',
});

Provides:

• Total token cost estimation
• Largest files by token count
• Optimization opportunities
• Cost projections at current API rates

Cache Performance

Monitor cache hit rates and storage efficiency:

// View cache statistics
await get_cache_stats({});

Metrics:

• Total entries
• Cache hit rate (%)
• Average compression ratio
• Total storage saved
• Most frequently accessed keys

Troubleshooting

CLI connection and discovery

The tools do not appear in Codex

1. Run codex mcp get token-optimizer.
2. Confirm the entry is enabled with codex mcp list.
3. Start a new Codex conversation.
4. Run /mcp in the interactive CLI and inspect the server status.

codex is blocked on Windows

PowerShell may reject the codex.ps1 shim under a restrictive execution policy. Use codex.cmd for the installation and verification commands, or review your user-scoped PowerShell execution policy.

Remove Token Optimizer from Codex

codex mcp remove token-optimizer

This removes the Codex registration; it does not delete your local cache or analytics database.

Common Issues and Solutions

Issue: "Invalid or malformed JSON" in Claude Code Settings

Symptom: Claude Code shows "Invalid Settings" error after running install-hooks

Cause: UTF-8 BOM (Byte Order Mark) was added to settings.json files

Solution: Upgrade to v3.0.2+ which fixes the BOM issue:

npm install -g @ooples/token-optimizer-mcp@latest

If you're already on v3.0.2+, manually remove the BOM:

# Windows: Remove BOM from settings.json
$content = Get-Content "~/.claude/settings.json" -Raw
$content = $content -replace '^\xEF\xBB\xBF', ''
$content | Set-Content "~/.claude/settings.json" -Encoding utf8NoBOM

# Linux: Remove BOM from settings.json
sed -i '1s/^\xEF\xBB\xBF//' ~/.claude/settings.json

# macOS: Remove BOM from settings.json (BSD sed requires empty string after -i)
sed -i '' '1s/^\xef\xbb\xbf//' ~/.claude/settings.json

Issue: Hooks Not Working After Installation

Symptom: Token optimization not occurring automatically

Diagnosis:

1. Check if hooks are installed:

   # Windows
   Get-Content ~/.claude/settings.json | ConvertFrom-Json | Select-Object -ExpandProperty hooks
   

   # macOS/Linux
   cat ~/.claude/settings.json | jq .hooks
   

2. Verify dispatcher.ps1 exists:

   # Windows
   Test-Path ~/.claude-global/hooks/dispatcher.ps1
   

   # macOS/Linux
   [ -f ~/.claude-global/hooks/dispatcher.sh ] && echo "Exists" || echo "Missing"
   

Solution: Re-run the installer:

# Windows
powershell -ExecutionPolicy Bypass -File install-hooks.ps1

# macOS/Linux
bash install-hooks.sh

Issue: Low Cache Hit Rate (<50%)

Symptom: Session stats show cache hit rate below 50%

Causes:

1. Working with many new files (expected)
2. Cache was recently cleared
3. TTL (time-to-live) is too short

Solutions:

1. Warm up the cache before starting work:

   await cache_warmup({
     paths: ['/path/to/frequently/used/files'],
     recursive: true,
   });
   

2. Increase TTL for stable APIs:

   await smart_api_fetch({
     url: 'https://api.example.com/data',
     ttl: 3600, // 1 hour instead of default 5 minutes
   });
   

3. Check cache size limits:

   await smart_cache({
     operation: 'configure',
     l1MaxSize: 2000, // Increase from default 1000
     l2MaxSize: 20000, // Increase from default 10000
   });
   

Issue: High Memory Usage

Symptom: Node.js process using excessive memory

Cause: Large cache in memory (L1/L2 tiers)

Solution: Configure cache limits:

await smart_cache({
  operation: 'configure',
  evictionStrategy: 'LRU', // Least Recently Used
  l1MaxSize: 500, // Reduce L1 cache
  l2MaxSize: 5000, // Reduce L2 cache
});

Or clear the cache:

await clear_cache({});

Issue: Slow First-Time Operations

Symptom: Initial Read/Grep/Glob operations are slow

Cause: Cache is empty, building indexes

Solution: This is expected behavior. Subsequent operations will be 80-90% faster.

To pre-warm the cache:

await cache_warmup({
  paths: ['/src', '/tests', '/docs'],
  recursive: true,
  schedule: 'startup', // Auto-warm on every session start
});

Issue: "Permission denied" Errors on Windows

Symptom: Cannot write to cache or log files

Cause: PowerShell execution policy or file permissions

Solution:

1. Set execution policy:

   Set-ExecutionPolicy -ExecutionPolicy RemoteSigned -Scope CurrentUser
   

2. Check file permissions:

   icacls "$env:USERPROFILE\.token-optimizer"
   

3. Re-run installer as Administrator if needed

Issue: Cache Files Growing Too Large

Symptom: ~/.token-optimizer/cache.db is >1GB

Cause: Caching very large files or many API responses

Solution:

1. Clear old entries:

   await clear_cache({ olderThan: 7 }); // Clear entries older than 7 days
   

2. Reduce cache retention:

   await smart_cache({
     operation: 'configure',
     defaultTTL: 3600, // 1 hour instead of 7 days
   });
   

3. Manually delete cache (nuclear option):

   rm -rf ~/.token-optimizer/cache.db
   

Getting Help

If you encounter issues not covered here:

1. Check the hook logs: ~/.claude-global/hooks/logs/dispatcher.log
2. Check session data: ~/.claude-global/hooks/data/current-session.txt
3. File an issue: GitHub Issues
   • Include debug logs
   • Include your OS and Node.js version
   • Include the output of get_session_stats

Limitations

• Small Text: Best for content >500 characters (cache overhead on small snippets)
• One-Time Content: No benefit for content that won't be referenced again
• Cache Storage: Automatic cleanup after 7 days to prevent disk usage issues
• Token Counting: Uses GPT-4 tokenizer (approximation for Claude, but close enough)

Development

git clone https://github.com/ooples/token-optimizer-mcp.git
cd token-optimizer-mcp
npm ci
npm run build
npm test
node scripts/mcp-smoke.mjs

To make Codex use your local build while developing:

codex mcp add token-optimizer-local -- node /absolute/path/to/token-optimizer-mcp/dist/server/index.js

Documentation

• Quick start
• Tool reference
• Codex and agent guidance
• Hook installation
• Testing
• Contributing
• Security policy
• Changelog

License

MIT License - see LICENSE for details

Author

Built for measurable token efficiency across supported coding agents by the ooples team.