What is an agent harness?
A model answers; an agent acts. An agent harness is the runtime that turns one into the other: the model thinks, the harness decides what that thinking is allowed to touch.
Simon Willison's definition of the agent itself is the cleanest: "an LLM agent runs tools in a loop to achieve a goal." The harness is everything around that loop: which tools exist, what needs approval, what the model sees each turn, what survives a crash. Andrej Karpathy named the architecture back in 2023: the model is "the kernel process of a new Operating System", and the harness is the rest of that OS, its scheduler, permissions, and memory. The SWE-agent paper proved the stakes by coining the agent-computer interface: how tools and feedback are presented changes what a model can do, independent of the model. The field's advice has since converged on investing here rather than in framework plumbing, from Anthropic's build-simple guidance to Jerry Liu's argument that the framework era is over and the layers that matter now are skills, tools, and context quality. Those are the layers this list catalogs.
Why harnesses matter
Better models make harnesses more important: more capabilities mean more failure modes, and production needs retry logic, fallbacks, and validation. Harness quality, not just model quality, determines whether agents actually ship. This list ranks projects by relevance to harness concerns (environment, orchestration, lifecycle, guardrails) and by stars/activity.
The benchmark data now backs this up. On SWE-bench Pro, "swapping the agent harness changed pass@1 more than many model upgrades do" (AINews, Aug 8 2026, citing analysis by @joelniklaus). Same model, different harness: 23% to 52% pass@1 on GLM-5.2, and 15% to 36% on Gemma 4 26B. Harness rankings barely transfer across models (rank correlation -0.05), so a small model in the right harness can approach a much larger model in the wrong one.
The gap is widest on long-horizon work. A bare frontier model was verified at about 30% on ARC-AGI-3; Prime Agent's harness took Opus 5 to 95.5%, and the YC Paper Club talk on why the harness matters more than the model (September 2026) walks through how. The gains come from a new class of self-improving harnesses, not from piling on scaffolding: the best of them stay thin and expose only what the model cannot do for itself. And because rankings barely transfer across models, the harness choice is a pairing that must be re-asked whenever the model changes. Who says so, what they measured, and what the claim does not mean is its own page.
That is the problem the MCP server in this repo solves. Point your agent at it and it can call recommend or pick_harness to choose a harness matched to your model and task, instead of inheriting whichever harness someone else benchmarked.
The landscape at a glance
Every project in the list, plotted by adoption surface area (the simplicity ↔ capability axis) against GitHub stars. Colors are categories; the largest projects in each tier are labeled.
The same projects placed by how much unsupervised rope they're designed to give (autonomy) and what happens when a run dies (recovery). In the tables below, ★ marks headless-ready projects and ✱ marks durable ones. Both charts regenerate from the list data on every refresh.
How to Pick a Harness
Start with the guide, then the head-to-head decision pages — grounded in the same data as the tables below:
- Why the harness matters more than the model: the measurements (ARC-AGI-3 30% to 95.5% on the same weights), who says so, and what the claim does not mean
- How to pick a harness: six questions that turn this list into a decision, plus the chart to internalize first (the harness moves scores more than the model)
- How to test-drive a harness: the two-week trial protocol, with a fair setup, tasks from your own repos, seven measurements, and the walk-away test
- The best AI agent harnesses in 2026, ranked by category: the top three per category from this week's data, regenerated on every refresh
- OpenClaw vs Hermes — the always-on personal-agent debate: presence vs discipline, plus what the field reports actually say
- Managed vs self-hosted always-on agents: Grok Bot, Claude Managed Agents, QM, OpenClaw, Hermes, and OpenJarvis; who owns the computer and who pays for idle time
- Terminal coding agents — opencode vs Codex vs Gemini CLI vs crush vs goose
- Multi-agent orchestration — OpenAI Agents SDK vs CrewAI vs AutoGen vs LangGraph
- — Mem0 vs Letta vs claude-mem
Templates and Playbooks
The list tells you which harness to use. These tell you how to set it up. Templates are files you copy into your project; Playbooks walk you through one task, step by step.
Templates
- One AGENTS.md for every coding agent: One briefing file that Codex, Claude Code, Cursor, OpenCode, GitHub Copilot, Gemini CLI, and Aider all read, so you write your build commands, conventions, and hard rules once instead of once per tool.
- Safe Claude Code settings: permissions and a guard hook: Project settings that stop Claude Code from force-pushing, wiping work, reading secrets, or piping downloads into a shell, while leaving everyday commands alone. Two files, copy and commit.
- Minimal agent harness in Python: A working coding-agent harness in about 180 lines of Python: the loop, three tools, a permission gate, a context file, a turn budget, and a transcript you can resume after a crash. Copy it to learn how harnesses work, or as the start of your own.
Playbooks
- Build your own agent harness: Build a working coding agent in about an hour: a loop, three tools, permissions, a context file, a budget, and crash recovery, in one Python file you fully understand. You finish with the minimal harness template running on your own repo.
- Write one AGENTS.md for every coding agent: Write one briefing file that Codex, Claude Code, Cursor, OpenCode, Copilot, Gemini CLI, and Aider all read, test that each tool actually loaded it, and keep it short enough to help instead of hurt. You finish with the AGENTS.md template filled in for your repo.
Skills
A skill is a folder your agent loads when a task calls for it. Each one below runs a script on your own setup and hands back a result you can check: what your agent can reach, which of its claims had no proof, where your tokens went. They run on your machine, and anything that changes a file or spends money waits for your yes.
Pick a harness
- Test-drive coding agents on tasks from your own repo: Runs Claude Code, Codex, and Gemini CLI on tasks taken from your own git history, scores each one with your repository's tests, and reports how many tasks each passed, the dollars per pass, the minutes, and the size of each change.
Set it up
- Check which instruction files your coding agents load: See which instruction files Claude Code, Codex, Gemini CLI, OpenCode, Cursor, GitHub Copilot, and Aider load from your repo, what each one cuts or skips, and whether the commands in those files still work.
- Grade your MCP tools the way a model reads them: Point it at an MCP server or a saved tool list to grade each tool's purpose, parameters, annotations, and token size, or let it find the MCP servers configured across your coding agents and count what they load into every session.
Keep it safe
- Test your agent's guardrails against dangerous commands: Checks whether the permission rules and hooks you already have in Claude Code, Codex, Gemini CLI, OpenCode, or Cursor stop about 90 dangerous commands, shows each one that gets through with a tested fix, and counts how often your rules interrupt real work.
- Stop a runaway agent session: A hook for Claude Code and Codex that stops a live session when the agent repeats the same call, keeps failing, or spends past a dollar cap you set.
- Check what your coding agent can reach: A one-minute probe that runs through your agent's own shell and shows which secret files, trust files, Docker, and sudo access it really has, then compares that with what your sandbox settings claim.
Check its work
- Check your agent's 'tests pass' claims: Find out how often your coding agent said tests passed without a passing run to back it up, and whether the current change quietly weakened your tests. It reads your local session logs and sends nothing anywhere.
- Turn the rules your agent breaks into hooks: Finds the rules in AGENTS.md, CLAUDE.md, and GEMINI.md that your coding agent breaks, counts every break in your recent sessions, and turns each broken rule into a hook (a check your agent runs before each tool call) that blocks it, tested against those real breaks.
See what it costs and what changed
- Find the update where your coding agent started working differently: Splits your own Claude Code or Codex sessions by version, model, or week, measures how the agent works in each, and places the change at the update, or within the span of versions, where the numbers moved, along with anything else that changed then.
- See where your coding agent wastes tokens and money: A report on your own Claude Code, Codex, Gemini CLI, and OpenCode sessions that shows which habits cost the most, such as re-read files, huge tool outputs, polling, and breaks that expire the prompt cache (the provider's cheap copy of the conversation), and which failures repeat, each with a fix.
Install one skill in any agent:
npx skills add https://github.com/RyanAlberts/best-of-Agent-Harnesses/tree/main/skills/<name>
Or all of them in Claude Code:
/plugin marketplace add RyanAlberts/best-of-Agent-Harnesses
/plugin install harness-skills@agent-harnesses
How the skills are built and tested.
Pick by use case
Reader's index: pick by what you want to do, not by category. Tag chips (e.g. mcp · memory) next to each row let you cross-filter by capability — see TAGS.md for the full cross-reference.
- I want a turnkey coding agent today — opencode, Cline, Codex, Gemini CLI, OpenHands, crush, Prime Agent · see Coding agent products (IDEs, CLIs, full suites)
- I want an always-on personal agent that lives in my chat apps — OpenClaw, Hermes, Khoj, Agent Zero, OpenHarness (HKUDS), QM, OpenJarvis · see Personal agent runtimes
- I want to extend Claude Code, Codex, or OpenCode with skills and slash commands — Anthropic Skills, , , , · see
For agents
This list is also published in machine-readable form, so coding agents and research agents can recommend harnesses — not just humans browsing GitHub:
- harnesses.json — every project with category, complexity tier, capability tags, stars, license signal, and a concrete example link, plus the full use-case index.
- llms.txt — the entire list in one agent-readable file. Point any agent at the raw URL.
- MCP server —
recommend (one opinionated pick + alternatives + what to avoid, e.g. repos flagged for star manipulation), compare/compare_for (2–4 harnesses side by side — by id or by task — who leads on which axis incl. researched sandboxing/memory/hooks/prompt-optimization ratings, graveyard warnings, the matching decision guide), pick_harness (ranked, with complexity/autonomy/recovery filters), pick_infrastructure (picks at any level of the infra stack plus a live GitHub/Hacker News discovery pass, so answers aren't limited to this list), search_harnesses, get_harness, list_categories, whats_new (the weekly digest), plus list_comparisons/get_comparison for the decision guides and list_templates/get_template/list_playbooks/get_playbook/list_skills/get_skill so your agent can install a template or a skill for you. Published to PyPI and the official MCP registry as io.github.RyanAlberts/agent-harnesses. One-line install (needs uv):
claude mcp add agent-harnesses -- uvx agent-harnesses-mcp
Or hire a skeleton
Don't just read the list — agents/ ships three agent skeletons: open-source agents that run on the AI subscription you already pay for. Clone the file, customize the instructions, done. All three work against the current week's data and deliver to Slack or Notion when either is connected:
- harness-scout — describe what you're building; it picks your harness, with evidence and a graveyard check.
- stack-auditor — flags the harnesses in your codebase that died, and can trace your agent session logs to show how the harness steers your technical decisions.
- harness-radar — weekly movement briefing: climbers, arrivals, deaths, graduations.
curl -fsSL https://raw.githubusercontent.com/RyanAlberts/best-of-Agent-Harnesses/main/agents/harness-scout.md -o .claude/agents/harness-scout.md
Contents
Guide to rankings
- ⭐ Stars — GitHub star count, captured 2026-10-25; tables sort by stars descending.
- ⚖️ Simplicity ↔ capability — adoption surface, 4 tiers: super simple (a format, one concept) → mostly simple (thin layer) → slightly complex (real SDK) → complex (product suite).
- ★ Headless-ready — designed for unattended runs, batches, and fleets (the top of the autonomy scale: step-gated → checkpoint-gated → bounded → headless).
- ✱ Durable — persisted execution state survives restarts mid-task (the top of the recovery scale: none → retry → resumable → durable).
- archived upstream — the maintainers archived the repo, so it gets no new fixes. Kept in the table for reference, never in the use-case picks.
- ✅ Open source — ✅ standard OSS license · ⚠️ source-available/restricted · ❓ no or unclear license.
- 🏷️ Tags — capability chips auto-derived from descriptions; full cross-reference in TAGS.md.
- 🎯 Examples — one concrete "show me it in action" link per project, not a docs root.
Every project's full autonomy and recovery tier is plotted in the grid above and carried in harnesses.json and llms.txt; scores are editorial, from public docs — maintainer corrections via issue/PR are merged fast.
Progressive disclosure harnesses
Formats, runtimes, and patterns that reveal context, tools, or instructions in layers—index first, details on demand—to control tokens and improve agent focus (the "map, not encyclopedia" principle).
| # | Project | ⭐ Stars | Description | Open source | Simplicity ↔ capability | Examples |
|---|
| 1 | Headroom | 77.8k | Compresses tool outputs, logs, files, and RAG chunks with content-aware compressors before they reach the model—claimed 20% fewer tokens for coding agents and 60–95% fewer for JSON, same answers. Ships as a library, HTTP proxy, or MCP server, so it drops in front of whatever harness you already run. mcp · rag | ✅ | mostly simple (compression library/proxy/MCP server) | Project README |
| 2 | awesome-cursorrules | 41k | Curated .cursorrules and skills that leverage Cursor's index-then-load model; the canonical collection for rules-as-progressive-disclosure in the IDE. ide | ✅ | super simple (content bundle) | PyTorch cursorrules |
| 3 | context-mode | 32.2k | Context-window optimization layer that sandboxes tool output before it reaches the model (claimed 98% reduction) and persists session memory across 17 agent platforms via MCP and hooks—progressive disclosure applied to tool results, not just instructions. mcp · memory · sandbox | ⚠️ Elastic-2.0 | mostly simple (output sandboxing, cross-platform) |
Coding agent products (IDEs, CLIs, full suites)
Turnkey coding agents you install and run: IDE extensions, terminal CLIs, Dockerized workspaces. Each entry notes which part is the harness (the agent loop, tool wiring, approval model) versus the UI shell (VS Code extension, TUI, browser client).
| # | Project | ⭐ Stars | Description | Open source | Simplicity ↔ capability | Examples |
|---|
| 1 | DeepSeek Harness | 261k | DeepSeek's official open-source agent harness (dsh), built so that everything is a plugin: the model adapter, tool registry, session log, sandbox and approval policy, and the agent loop itself are Cordis plugins you replace from configuration, and one plugin tree boots as a Web UI, a headless runner, a desktop app, an SDK server, or an ACP server. Developer preview, with breaking changes expected. browser · sandbox · typescript | ✅ | slightly complex (plugin-tree runtime, Web UI, headless, SDK) | Plugin architecture |
| 2 | opencode ★ | 221k | Open-source terminal coding agent (formerly sst/opencode; transferred to anomalyco). The harness is a multi-provider tool-call loop (Claude, OpenAI, Gemini, local) with strong plugin and MCP support; the TUI is the shell. 100% OSS, very actively shipped. mcp · provider-agnostic · cli · tui · typescript | ✅ | slightly complex (multi-provider, plugins, MCP) | Agent system page |
| 3 | Codex | 131k |
Coding harness configs and SDKs
Skill packs, slash-command libraries, meta-prompting frameworks, and official SDKs that give you the harness (the agent loop, planning, memory, hooks) without bundling a specific IDE or CLI shell.
| # | Project | ⭐ Stars | Description | Open source | Simplicity ↔ capability | Examples |
|---|
| 1 | superpowers | 305k | Performance-oriented harness pack for Claude Code and 13 other harnesses (Codex, Cursor, OpenCode, Gemini CLI, more): skills, instincts, memory, security, research-first workflows. Treats harness engineering itself as the performance lever. memory · cli · ide | ✅ | complex (multi-IDE skill stack — product suite) | TDD skill |
| 2 | Anthropic Skills | 184k | Anthropic's official Agent Skills repository: SKILL.md-based folders (instructions, scripts, resources) Claude dynamically loads on Claude Code, Claude.ai, and the API. The reference for progressive-disclosure skill packs in 2026. | ⚠️ Anthropic terms | mostly simple (official skills format) | docx skill |
| 3 | GStack | 138k | Garry Tan's Claude Code skill stack: 23 slash-command modes (CEO/eng/design review, QA, ship, browse, retro, …) that structure one assistant as a virtual engineering team. Daily driver while running YC. typescript | ✅ | slightly complex (multi-role slash-command harness) | /ship SKILL.md |
Personal agent runtimes
Always-on, self-hosted agents you run as a daemon and talk to from chat apps: gateway runtimes, second brains, and self-improving assistants. The agent as a product you operate, not a library you build with.
| # | Project | ⭐ Stars | Description | Open source | Simplicity ↔ capability | Examples |
|---|
| 1 | OpenClaw ★ | 393k | Self-hosted, always-on personal agent (formerly Clawdbot/Moltbot): a gateway + event-loop runtime that treats messages, heartbeats, crons, and webhooks as one input queue, persists state to local files, and lives in your chat apps (WhatsApp, Telegram, Slack, Discord). 13,700+ community skills; the fastest-growing repo in GitHub history. typescript · multi-agent | ✅ | complex (always-on runtime, channels, skill ecosystem — product suite) | Agent runtime architecture |
| 2 | Hermes ★ | 257k | Nous Research's self-improving agent: a learning loop turns experience into reusable skills, builds a persistent user model across sessions, and checkpoints state to disk with rollback; lean enough for a $5 VPS, driven from chat, and model-agnostic (Nous Portal, OpenRouter, OpenAI, or any endpoint). memory · python · provider-agnostic | ✅ | slightly complex (lean runtime, learning loop, disk-first memory) | Built-in skills |
| 3 | AnythingLLM ★ | 67.5k | Self-hosted "AI second brain" harness: chat with your documents, run built-in agent skills (web search, code execution, browsing), and manage multi-user workspaces with a bundled vector DB, so there is no separate RAG stack to wire up. · |
Frameworks
General-purpose agent and LLM application frameworks (the app layer, not harnesses per se).
| # | Project | ⭐ Stars | Description | Open source | Simplicity ↔ capability | Examples |
|---|
| 1 | n8n ★ ✱ | 209k | Fair-code workflow engine with 400+ nodes and native AI nodes; the self-hosted Zapier that actually does agents and LangChain. workflow · local · typescript | ⚠️ Fair-code | complex (400+ nodes, workflow engine — product suite) | Agent vs chain workflow |
| 2 | AutoGPT ★ | 188k | The original autonomous loop: goal in, agent iterates with tools and memory; Forge is the dev framework, Benchmark the eval harness. memory · evals · python | ⚠️ Polyform-SU | complex (autonomous loop, tools, memory — product suite) | Medium blogger graph |
| 3 | Dify ★ | 160k | One-stop LLM app platform: visual workflows, RAG pipeline, 50+ tools, model management; "ship from prototype to prod" in a single UI. low-code · rag · python | ⚠️ Fair-code | complex (one-stop platform — product suite) | Customer-service bot |
Multi-agent and orchestration
Harnesses and patterns for multi-agent coordination and handoffs.
| # | Project | ⭐ Stars | Description | Open source | Simplicity ↔ capability | Examples |
|---|
| 1 | MetaGPT ★ | 71k | The "AI software company" multi-agent framework: role-played PM, architect, and engineer agents turn a one-line requirement into specs, designs, and code along an SOP assembly line. The landmark of the genre; development pace has slowed in 2026. multi-agent · python | ✅ | complex (role pipeline, SOPs — product suite) | Build a customized agent |
| 2 | autogen | 61.6k | Conversable agents and group chats; code execution and human-in-the-loop; Microsoft origin, AG2 ecosystem. ⚠️ In maintenance mode since late 2025: no new features, and Microsoft directs new users to Agent Framework. multi-agent · python | ✅ CC-BY | complex (group chat, code exec, AG2 — product suite) | Distributed group chat |
| 3 | crewAI | 60.1k | Role-based agents (roles, goals, backstories) in Crews; Flows add event-driven and hierarchical control for production. python | ✅ | complex (roles, Flows, production — product suite) | Trip planner crew |
Plugins, MCPs, CLI tools
IDE plugins, concrete MCP servers, and CLI tools that give agents tools and context.
| # | Project | ⭐ Stars | Description | Open source | Simplicity ↔ capability | Examples |
|---|
| 1 | MCP Servers | 91.6k | The official reference collection of Model Context Protocol servers (filesystem, git, fetch, memory, time, and more)—the canonical, vetted toolset agents connect to, and the pattern every other MCP server is measured against. mcp · memory · typescript | ✅ | mostly simple (reference servers) | Server catalog |
| 2 | Context7 | 63.7k | MCP server that injects up-to-date, version-specific library docs into an agent's context on demand; kills the stale-training-data hallucinations that plague codegen. mcp · training · typescript | ✅ | super simple (drop-in MCP) | Docs |
| 3 | chrome-devtools-mcp | 54.1k | Google's official Chrome DevTools MCP server: exposes console, network, and performance-trace inspection as tool calls agents can drive directly, instead of a human clicking through the DevTools panel. mcp · browser · typescript | ✅ | mostly simple (official browser-debug MCP) | Project README |
Memory and state
Persistent memory layers that give agents recall across turns and sessions: knowledge graphs, vector stores, and session-capture tools that survive a restart. The state a harness needs but rarely ships with.
| # | Project | ⭐ Stars | Description | Open source | Simplicity ↔ capability | Examples |
|---|
| 1 | claude-mem | 103k | Session-memory plugin for Claude Code, Codex, OpenClaw, Gemini, Copilot, and more: captures everything an agent does during a session, AI-compresses it, and injects the relevant context into future sessions. Session-to-session memory as a drop-in. memory | ✅ | slightly complex (session capture + compression) | Lifecycle hooks config |
| 2 | Mem0 | 67.8k | Universal memory layer for AI agents: stores user/org/session memory, retrieves on demand. Apache-2.0; the de-facto memory primitive paired with most harnesses in 2026. memory · python | ✅ | slightly complex (memory layer, multi-platform) | Next.js memory demo |
| 3 | cognee | 32.7k | Open-source memory layer for agents: an extract–cognify–load pipeline that turns your data into a queryable knowledge graph plus vector store, so agents recall facts and relationships across sessions instead of re-reading context. memory · rag · workflow · python | ✅ | slightly complex (graph + vector memory) | Quickstart |
Evaluation and benchmarking harnesses
Agentic eval systems, reasoning benchmarks, and open agent benchmarks.
| # | Project | ⭐ Stars | Description | Open source | Simplicity ↔ capability | Examples |
|---|
| 1 | Agent Lightning ★ | 18.8k | Microsoft's training-oriented harness: optimization loops for agent behavior—when you need to improve policies over rollouts, not only score a fixed prompt. evals · training · python | ✅ | complex (agent training, Microsoft stack — product suite) | APO room-booking example |
| 2 | SWE-bench ★ | 6.1k | LMs resolve real GitHub issues; Docker harness, instance IDs; standard for code-agent evals. evals · sandbox · python | ✅ | slightly complex (real GitHub issues, standard) | SWE-bench Verified leaderboard |
| 3 | AgentBench ★ | 3.8k | ICLR'24 benchmark: agents across AlfWorld, DB, knowledge graphs, OS, webshop; Docker Compose, function-calling interface. evals · sandbox · rag · workflow · python | ✅ | complex (multi-env, Docker Compose — product suite) | AgentBench ICLR'24 paper |
Observability and eval-ops
Tracing, monitoring, and production evaluation for live agent runs: capture every step, tool call, and token, then score and debug in the loop. Distinct from the fixed-task benchmarks above—this is what you run against your own traffic.
| # | Project | ⭐ Stars | Description | Open source | Simplicity ↔ capability | Examples |
|---|
| 1 | Langfuse | 36.5k | Open-source LLM engineering platform: full-trace observability, online and offline evals, prompt management, and cost metrics for agent runs in production—the monitoring layer most harnesses lack out of the box. evals · typescript | ✅ | slightly complex (tracing + evals platform) | Docs |
| 2 | MLflow | 28.5k | Mature ML platform now covering GenAI: MLflow Tracing captures every agent step, tool call, and token, with built-in LLM evals and prompt versioning—observability for teams already standardized on MLflow. evals · python | ✅ | complex (full ML + GenAI platform) | Docs |
| 3 | Opik | 22.9k | Comet's open-source agent observability and evaluation platform: tracing, scoring, and experiment comparison with the whole core feature set free to self-host under Apache-2.0. evals · python | ✅ | slightly complex (tracing + evals platform) | Docs |
| 4 | |
Research and task-specific harnesses
Deep research, document QA, and domain-specific agent loops.
| # | Project | ⭐ Stars | Description | Open source | Simplicity ↔ capability | Examples |
|---|
| 1 | DeerFlow ★ | 84.4k | ByteDance's long-horizon research harness: a LangGraph-based agent loop that researches, codes, and creates using sandboxes, memory, tools, and subagents behind a message gateway—an alternative to the reference gpt-researcher stack for multi-hour tasks. memory · multi-agent · sandbox · python | ✅ | complex (sandboxes, memory, subagents — product suite) | Project README |
| 2 | gpt-researcher | 30.3k | Autonomous deep-research agent: web + local sources, citation-grounded reports, multi-agent and deep-research modes. The reference open-source research harness. multi-agent · python | ✅ | complex (deep research, multi-agent — product suite) | Multi-agent LangGraph walkthrough |
| 3 | AutoResearchClaw | 14.7k | Fully autonomous, self-evolving research harness: multi-agent debate and citation verification carry a project from idea to a written paper without a human in the loop. multi-agent | ❓ | complex (autonomous research, multi-agent debate — product suite) |
Libraries and SDKs
Lightweight runtimes, tool loops, and provider-agnostic harness primitives.
| # | Project | ⭐ Stars | Description | Open source | Simplicity ↔ capability | Examples |
|---|
| 1 | LiteLLM | 62k | One interface to 100+ LLMs; routing, caching, budgets. Not an agent framework—the pipe every agent framework uses. provider-agnostic · python | ✅ | mostly simple (LLM pipe only) | Anthropic Agent SDK gateway |
| 2 | Composio | 30.7k | 1,000+ toolkits with auth, tool search, and a sandboxed workbench—drop-in tool layer so agents stop reinventing OAuth + integrations. Python and TypeScript. sandbox · tool-discovery · python · typescript | ✅ | complex (1k+ tools, auth, search — product suite) | HackerNews agent quickstart |
| 3 | deepagents ✱ | 30.6k | LangChain's Python+TypeScript agent harness on top of LangGraph: planning tool, virtual filesystem, shell sandbox, sub-agent spawning—the "Claude Code-style" harness as a reusable library. multi-agent · sandbox · python · typescript | ✅ | slightly complex (planning, files, sub-agents) | Deep research agent |
⚰️ Graveyard
Archived upstream, or flagged for curation integrity (e.g. suspected star manipulation). Kept here — not deleted — for citation and transparency; excluded from the ranked count, the landscape chart, and harnesses.json's main list. Curation is the point: a starred repo is not automatically a credible one.
| Project | Last ⭐ Stars | Since | Why it's here |
|---|
| everything-claude-code | 289k | 2026-07-11 | suspected star manipulation — ~228k stars / ~35k forks on a repo created 2026-01 with no matching install base, dependents, or discussion; fork-to-star ratio and growth curve are inconsistent with organic adoption |
| Daytona | 71.5k | 2026-10-04 | archived upstream — kept for citation |
| spring-ai-tool-search-tool | 77 | 2026-07-03 | archived upstream — kept for citation |
| coderClaw | 3 | 2026-07-03 | archived upstream — kept for citation |
🔭 On the radar
Up-and-coming candidates — surfaced by the weekly discovery scan or submitted by the community — that haven't cleared the curation bar or a vetting pass yet. Stars refresh weekly from the discovery queue; descriptions are the projects' own, unvetted. Entries graduate into the ranked list above or drop off.
| Project | ⭐ Stars | What it says it is | Via |
|---|
| agenticSeek | 27.5k | Fully Local Manus AI. No APIs, No $200 monthly bills. Enjoy an autonomous agent that thinks, browses the web, and code for the sole cost of electricity. | weekly discovery |
| openfang | 18.2k | Open-source Agent Operating System | weekly discovery |
| evolver | 8.7k | The GEP-powered self-evolving engine for AI agents. Auditable evolution with Genes, Capsules, and Events. | evomap.ai | weekly discovery |
| ZCode | 8.2k | Z.ai's coding agent harness. Powerful, intelligent, extensible. | weekly discovery |
| reef | 7.8k | Infrastructure for continually self-improving agents: records live traffic and feedback, then ships a scored update to model weights or the harness. | community · PR #125 |
| openevolve | |
FAQ
Which agent harnesses can run unattended (headless)?
Harnesses designed for unattended runs, batches, and fleets: opencode, OpenHands, goose, Symphony, Prime Agent, SWE-agent, Claude Agent SDK, Ouroboros (Q00).
Which agent harnesses survive a crash mid-task (durable)?
Harnesses whose execution state persists across restarts: langgraph-bigtool, rpi, Ouroboros (Q00), QM, n8n, langgraph, mastra, letta.
How many of these agent harnesses are open source?
142 of 187 carry a standard open-source license; the rest are source-available or unclear, and flagged per row.
What is an agent harness?
The runtime that turns a model into an agent: it decides what the model's reasoning is allowed to touch, and supplies the orchestration, tool wiring, memory, error recovery, and guardrails around per-turn inference.
Does the harness matter more than the model?
Often, yes, and measurably: the same weights scored about 30% on ARC-AGI-3 as a bare model and 95.5% inside the Prime Agent harness, and on SWE-bench Pro swapping only the harness moved GLM-5.2 from 23% to 52%. Harness rankings barely transfer across models (rank correlation about -0.05), so pick the harness and the model as a pair, and re-pick when the model changes. Sources and the full argument
Is Grok Bot an agent harness?
Yes, a managed one: xAI owns the loop, the tool wiring, the memory, and the approval rules, and every Bot on an account shares one cloud computer. It is not in the ranked list because the list ranks open repositories; the managed-vs-self-hosted guide compares it with Claude Managed Agents, QM, OpenClaw, Hermes, and OpenJarvis. Sources and the full argument
How is this list ranked?
By relevance to harness concerns (environment, orchestration, lifecycle, guardrails) and by GitHub stars (captured 2026-10-25); each project also carries an adoption-surface tier and autonomy/recovery scores.
How can an AI agent use this list directly?
Three machine-readable surfaces: harnesses.json (structured), llms.txt (one file), and an MCP server (uvx agent-harnesses-mcp) exposing recommend, compare, pick_harness, and search_harnesses.
Related Resources
- Awesome: Awesome lists on many topics
- OpenAI – Harness engineering: Environment design, intent, feedback loops, repo-as-system-of-record
- Anthropic – Effective harnesses for long-running agents: Session bridging, feature lists, incremental progress, testing
- Aakash Gupta (Medium) – 2026 is agent harnesses: Harness as moat, minimal intervention, progressive disclosure
- YC Paper Club (video) – Why the harness matters more than the model: September 2026 session with the authors of Prime Agent, OpenJarvis, and QM; the history of harnesses runs from 7:00 to 17:00
- learn-agent-architecture: Section-by-section teardown of how Claude Code and Hermes Agent structure the agent loop, tools, permissions, context, and memory, with a small runnable Python demo per mechanism
- LangChain, Anthropic, OpenAI: Official docs for major agent platforms
🧡 Thank you, contributors
The people who stopped mid-scroll, found a gap, and wrote it up — this list is better for each of them:
@baskduf — harness-starter-kit · @ahwurm — LocalHarness · @liviux — LoopTroop · @rishabhpoddar — TeamCopilot, on the radar · @abilliontokens — oh-my-pi · @pranshuchittora — agent-qa · @madarco — AgentBox · @jmthomasofficial — JMT x402 Agent Tools · @ShukantPal — Proliferate · @hjqcan — GoodMemory · @777genius — Agent Teams AI · @S1LV3RJ1NX — mcp-guardian · @AmariahAK — Atlarix, Update Atlarix entry · @hardness1020 — awesome-agent-architecture · @razzant — Claudexor, Ouroboros · @allenshi16 — Nexus AI Pulse, HomeOffice AI · — fractal · — ClawBench evaluation harness · — Pin mcp2 · — L∞pGate · — red-team-blue-team-agent-fabric · — Open Multi-Agent · — Ouroboros · — PR #78 · — LWC Local Wiki CLI · — MCP Lens · — Prime Agent · — SandBase Harness · — deja, deja-vu · — MCP spec conformance · — HarnessRouter · — SandBase Harness · — Orkas · — Nausicaa · — OrcaReplay · — YYLO, YYLO Benchmark · — Gram · — GraphCode · — Minimal Harness · — Model Citizen · — Mnemoverse · — Jev Social, Jet Browser · — Raven · — House Party Protocol · — Reef · — AutoBot · — Orbi, Orbi Coding agent products · — 5dive · — Tianshu Harness · — Desplega Agent Swarm · — Tale · — Empryo · — workkit · — Project suggestion · — Giovanni · — laravel-ai-verify · — rpi embeddable Rust coding-agent harness · — PlaceCall · — Browser Use Cloud · — AgentLink · — Sandboxed code execution · — wippaAGENT · — Scalekit
Accepted submissions land with co-author credit on the commit that ships them. Promising projects that are still early aren't turned away — they get pinned to 🔭 On the radar and graduate as they grow. Add yours →
Contribution
Contributions are welcome. To add or suggest projects:
- Open an issue with the repo URL, category, and a short description.
- Or submit a pull request against scripts/generate.py — this README, projects.yaml, and TAGS.md are generated from it, so direct edits to them can't merge.
Promising projects that don't clear the curation bar yet get pinned to 🔭 On the radar — a submission that lands there isn't rejected, it's queued.
For contribution guidelines, see CONTRIBUTING.md and the Code of Conduct.
Show your listing
If your project is in this list, you're welcome to show it in your README:
[](https://github.com/RyanAlberts/best-of-Agent-Harnesses)
License
The skills in skills/ are MIT-licensed, so you can copy them into any project.
