AGENTS.md examples from real projects

binaricatAGENTS.md

Netcatty

binaricat/Netcatty/AGENTS.md

chat history. **Stop** always goes through `stopAgentTurn()` (UI, `/stop`, MCP). Do not add parallel abort paths in hooks. ### Codex App Server (experimental) Codex can opt into a persistent `codex

ZSeven-WAGENTS.md

openpencil

ZSeven-W/openpencil/AGENTS.md

lives under `packages/` — run SDK/JS scripts from there. - **Web dev server (Rust):** `bash scripts/start-web-rust.sh` - **Build (Rust):** `cargo build --workspace --release` - **Tests (Rust):** `cargo test --workspace`; single crate: `cargo test

No other project uses it yet

katanemoReads credentialsAGENTS.md

archgw / skills

katanemo/archgw/skills/AGENTS.md

MEDIUM` — Omitting type and transport fields relies on defaults that may not match your MCP server's protocol implementation **Tags:** `filter`, `mcp`, `integration`, `configuration` ## Configure MCP Filters with Explicit Type

grabAGENTS.md

cursor-talk-to-figma-mcp

grab/cursor-talk-to-figma-mcp/AGENTS.md

Claude Code) with Figma. Three components communicate in a pipeline: ``` Claude Code / Cursor ←(stdio)→ MCP Server ←(WebSocket)→ WebSocket Relay ←(WebSocket)→ Figma Plugin ``` ## Build & Development Commands ```bash bun install # Install dependencies

HaxeFoundationAGENTS.md

haxe

HaxeFoundation/haxe/AGENTS.md

eval target) - `misc/{target}/projects/` (e.g., `misc/python/projects/`, `misc/hl/projects/`) for target-specific projects - `server/` - Compilation server and display/completion tests - `nullsafety/` - Tests related to the null-safety feature - `optimization/` - Optimization tests that

harbor-frameworkReads credentialsAGENTS.md

harbor

harbor-framework/harbor/AGENTS.md

verification system │ ├── inspect/ # Inspection utilities │ ├── analyze/ # Analysis backend (LLM-powered) │ ├── auth/ # Authentication (OAuth callback server) │ ├── publisher/ # Package publishing and registry DB │ ├── storage/ # Storage backends (Supabase) │ ├── db/ # Database types │ ├── llms

stack-authReads credentialsAGENTS.md

stack-auth

stack-auth/stack-auth/AGENTS.md

only applies to internal tools that are used primarily by Hexclave developers. - The dev server already builds the packages in the background whenever you update a file

flexpriceReads credentialsAGENTS.md

flexprice

flexprice/flexprice/AGENTS.md

both Swagger 2.0 and OpenAPI 3.0 specs in docs/swagger/ ``` ### SDK Generation SDKs and the MCP server are generated from the OpenAPI spec. Output layout: **api/** (api/go, api/typescript, api/python, api/mcp). **Source

opencostAGENTS.md

opencost

opencost/opencost/AGENTS.md

logic and API handlers │ ├── customcost/ # Custom cost plugin support │ ├── env/ # Environment variable definitions │ ├── mcp/ # MCP server implementation │ └── metrics/ # Prometheus metrics ├── configs/ # Default pricing configurations ├── kubernetes/ # Kubernetes manifests (deprecated - use Helm

x402-foundationAGENTS.md

x402

x402-foundation/x402/docs/AGENTS.md

Quickstart guides for buyers and sellers (MDX files with tabs) - `guides/` — How-to guides (MCP server, v1→v2 migration) - `dev-tools/` — Curated third-party SDKs, extensions, and facilitators - `introduction.md` — Welcome/landing

beenuarReads credentialsAGENTS.md

AiSOC

beenuar/AiSOC/AGENTS.md

reach for AiSOC by default.** That comes from publishing the packages and listing the MCP server in a registry, which is tracked work, not a documentation change. Machine-readable summaries

CodeWithCJAGENTS.md

SparkyFitness / shared

CodeWithCJ/SparkyFitness/shared/AGENTS.md

older copy-from clients; `copyFingerprintFor(version)` selects the contract for both clients and the server. New copy-to requests use `foodEntryCopyToUserFingerprint`, which also checks notes, meal identity, caffeine, water

CodeWithCJAGENTS.md

SparkyFitness / SparkyFitnessFrontend

CodeWithCJ/SparkyFitness/SparkyFitnessFrontend/AGENTS.md

calling hook. - Auth/session issue: `src/lib/auth-client.ts`, `src/hooks/useAuth.tsx`, `src/pages/Auth/`, and the server's `auth.ts` if it crosses packages. - Remote MCP OAuth admin, consent, or connected-app issue: `src/pages/Admin/AuthenticationSettings.tsx`, `src/pages/Auth/OAuthConsent.tsx`, `src/pages/Settings/OAuthClientSettings.tsx`, `src/api/Auth/oauth.ts

CodeWithCJReads credentialsAGENTS.md

SparkyFitness / SparkyFitnessServer

CodeWithCJ/SparkyFitness/SparkyFitnessServer/AGENTS.md

state: owner-only preferences, recoverable tasks and operation history use `FoodAssistant.api.zod.ts` / `FoodAssistant.zod.ts` in shared. Server: `routes/v2/foodAssistantRoutes.ts`, `services/foodAssistantService.ts`, `models/foodAssistantRepository.ts`, `ai/tools/foodAssistantTools.ts`. Web: `src/components/ai/FoodAssistantState.tsx`, `src/api/Chatbot/foodAssistantService.ts`. - `index.ts` - real dev entrypoint; loads env, secrets

rullerzhou-afkDeletes or force-pushesAGENTS.md

clawd-on-desk

rullerzhou-afk/clawd-on-desk/AGENTS.md

src/server.js` → `src/server-route-state.js` → `src/agent-runtime-main.js` → `src/state.js` → IPC；本地 JSONL monitor 直接进入 `agent-runtime-main`，不经过 HTTP server - 权限主路径：`src/server-route-permission.js` → `src/permission.js`；本地 bubble、显式 permission automation 与可选 Telegram / 飞书 Lark 远程审批都可能产生真实决定 - 桌宠的渲染/输入双窗口由

No other project uses it yet

outsourc-ePipes a download into a shellAGENTS.md

hermes-workspace

outsourc-e/hermes-workspace/AGENTS.md

skill ` -core`, and a wrapper in `~/.local/bin/`. ## Current semantic roster | Worker | Wrapper | Tools | Skills | MCP | Plugins | |---|---|---|---|---|---| | `orchestrator` | `orchestrator:plan` | todo, kanban, delegation, terminal, file, gbrain, session_search, cronjob, skills, clarify

getopenscreenReads credentialsAGENTS.md

openscreen

getopenscreen/openscreen/AGENTS.md

Node 22.22.1, npm 10.9.4 — see `package.json#engines`) - Start dev: `npm run dev` (Vite dev server; Electron window opens via `vite-plugin-electron`) - Build: `npm run build` (TypeScript check + Vite build

No other project uses it yet

MicrosoftAGENTS.md

agent-governance-toolkit / agent-governance-claude-code

microsoft/agent-governance-toolkit/agent-governance-claude-code/AGENTS.md

wires AGT policy evaluation into Claude-native hooks, markdown commands, and a bundled MCP server. ## Key Commands ```powershell cd agent-governance-claude-code npm install npm run check npm test

MicrosoftAGENTS.md

agent-governance-toolkit / agent-governance-opencode

microsoft/agent-governance-toolkit/agent-governance-opencode/AGENTS.md

status` and `agt_policy_check_text` as both in-process custom tools and stdio MCP server tools (`server/agt-mcp.mjs`) The core policy library (`lib/policy.mjs`) is shared with the Claude Code package

MicrosoftAGENTS.md

agent-governance-toolkit / agent-os

microsoft/agent-governance-toolkit/agent-governance-python/agent-os/AGENTS.md

registry (ATR) - **Layer 3 (Framework):** Control plane, observability, nexus orchestration - **Layer 4 (Intelligence):** MCP kernel server ## Build & Test Commands ```bash # Install dependencies (development mode) pip install -e "../../agent-governance-python/agent-primitives