LAMBLE — trenches research, powered by Frames

LAMBLE publishes research on the crypto trenches: coin pages with holder analysis and a bottoming test, narratives with their full membership and lifecycle history, launchpad economics, and a weekly State of the Trenches edition. People read it on the website; agents read the same research through a structured API, with fixed-price bundles paid per request (x402 or MPP). There are no reader accounts or subscriptions. Data is acquired through Frames; the launch bot remains available as a secondary link.

Status, requirement by requirement, with tests and evidence: docs/implementation/lamble-trenches-status.md.

How it fits together

Frames (paid reads, budget-guarded)        scripts/frames-refresh (plan, discover, normalize)
        │                                   scripts/lamble collect (programmatic) or an agent session
        ▼
research/lamble/<run>/  ── retained raw responses, requests, evidence, normalized outputs (per daily run)
        │  npm run lamble -- ingest
        ▼
data/lamble.sqlite       ── observation store: tokens, pools, observations (with availability time and
        │                  revisions), memberships, lifecycle, evidence (gitignored, rebuilt from runs)
        │  npm run lamble -- release / weekly / research
        ▼
data/releases/<id>/     ── immutable, content-addressed release: public.json (website), bundles.json.gz
data/editions/<week>/   ── weekly editions, revised (r2, r3…) never edited
data/changes/           ── one change feed: release diffs, corrections, editions, research
content/research/       ── articles generated from a release; every citation and chart checked before writing
        │
        ├── website (src/app)   reads public.json only
        └── /api/v1             free discovery + paid bundles (rights-gated, x402/MPP)

Serving a page or an API request never calls a provider. Builds read committed release files.

Commands

npm run dev                 # http://localhost:3000
npm run build && npm start
npm run lint
npm run typecheck
npm test                    # unit and integration tests (offline)
npm run data:setup          # rebuild the store and outputs from retained runs (offline, never spends)

npm run lamble -- providers validate|matrix|inspect|health [--plan|--run]   # catalog; health --run: free probes
npm run lamble -- ingest | release | weekly | research | refresh | status
npm run lamble -- collect research/lamble/<run> <batch…> [--dry-run]   # paid Frames reads, budget + catalog limits
npm run lamble -- deliver [--dry-run]                                  # change notifications (disabled)
npm run lamble -- deliver verify <id> | confirm <id> <code>           # verify a webhook or Telegram destination

Node 22.13 or later (built-in node:sqlite, --experimental-strip-types).

Configuration

Variable	Purpose	Default
NEXT_PUBLIC_SITE_URL	Canonical origin for metadata, feeds and payment resource URLs	http://localhost:3000
NEXT_PUBLIC_FRAMES_URL	Link target for "Powered by Frames"	https://frames.ag
NEXT_PUBLIC_FRAMES_MCP_URL	Frames MCP URL shown in agent setup	unset: readers are told to use their Frames account
NEXT_PUBLIC_LAMBLE_FRAMES_LISTING	listed only after the API is found in Frames search	unset: pages say "not yet listed"
NEXT_PUBLIC_LAMBLE_COVERAGE_REQUEST_URL	Intake for coverage requests	unset: copyable request text only
LAMBLE_X402_*, LAMBLE_MPP_*, MPP_SECRET_KEY	Payment protocols	disabled — see docs/runbooks/payments.md
LAMBLE_COMMERCE_DB_PATH	Commerce database (purchases, entitlements)	data/commerce.sqlite
LAMBLE_FRAMES_BUDGET_ID, LAMBLE_FRAMES_BUDGET_USD	Owner-approved budget for collect	unset: no paid call
FRAMES_MCP_URL, FRAMES_MCP_AUTHORIZATION	Frames MCP endpoint for collect and providers health --run	unset
LAMBLE_DELIVERY_ENABLED	Allow deliver to send	unset
LAMBLE_HIDE_PENDING_REVIEW	Hide unreviewed deep dives outside Vercel production	unset

Documentation

• Methods: docs/methods (the files the code reads; rendered at /methodology)
• Provider catalog: docs/data/providers.json, jobs and budgets: docs/data/jobs.json
• Runbooks: operations, provider onboarding, editorial policy, payments, Frames listing
• Daily acquisition routine: docs/prompts/refresh-market-data.md

Stack

Next.js 16 (App Router), React 19, TypeScript strict; Tailwind v4 bridged to the design tokens in src/styles/tokens.css (dark default); shadcn (Base UI) primitives in src/components/ui; lightweight-charts and inline SVG sparklines; @x402/* and mppx for payments; node:sqlite stores.

src/app                  pages, /api/v1 route handlers, feeds, sitemap, share images
src/components/research  research UI kit (measures with states, panels, evidence, charts, agent access)
src/components/launchpads  screener table and detail sheet
src/lib/lamble            core (identity, time, measures), catalog, adapters, methods, release types,
                         service (the one read path), api, payments, research, notify, frames
scripts/lamble            CLI: providers, ingest, release, weekly, research, collect, deliver, status
scripts/frames-refresh   daily acquisition planning and normalization (Python)