How real projects brief Codex, Cursor and every other agent that reads AGENTS.md.
AbilityaiAGENTS.mdAbilityai/trinity/AGENTS.md
detail read on demand: component/API catalogs, execution, reliability, agent lifecycle + runtime, integrations, observability, workspace, frontend, MCP server, DB schema, security | | [docs/TRINITY_COMPATIBLE_AGENT_GUIDE.md](docs/TRINITY_COMPATIBLE_AGENT_GUIDE.md) | Agent template structure in depth | | [docs/MULTI_AGENT_SYSTEM_GUIDE.md](docs/MULTI_AGENT_SYSTEM_GUIDE.md) | Multi
wuxiranAGENTS.mdwuxiran/cc-pane/AGENTS.md
project ## Technology Stack | Layer | Technology | Purpose | |-------|-----------|---------| | Desktop Framework | Tauri 2 | Rust backend + system WebView | | Frontend | React 19 + TypeScript 5.6 | UI components | | State Management | Zustand 5 + Immer | Immutable state updates
avibe-botAGENTS.mdavibe-bot/avibe/AGENTS.md
text only when required for localization/i18n - backend user-facing strings must go through `vibe/i18n/` - frontend user-facing strings must go through `ui/src/i18n/en.json` and `ui/src/i18n/zh.json` - never hardcode user-visible display text
bex-coAGENTS.mdbex-co/bex/dashboard/AGENTS.md
fidelity local bex needs the mock cluster - Ory stack (`scripts/mock-cluster.sh`), which is heavy for frontend work. `scripts/local-bex.mjs` (run via `yarn local-bex`, wired by `yarn dev:local`) is a tiny
miantiao-meAGENTS.mdmiantiao-me/aigc-weekly/AGENTS.md
value) || '格式不正确' // 可选链安全访问 doc.tags?.map(t => t.value) ?? [] ``` ## 架构概览 ``` aigc-weekly/ ├── app/ # Next.js App Router │ ├── (frontend)/ # 前台页面 │ └── (payload)/ # Payload Admin ├── agent/ # Claude Agent（运行在 Cloudflare Containers） ├── worker/ # Cloudflare Worker ├── collections/ # Payload
mx-spaceAGENTS.mdmx-space/core/AGENTS.md
apps/admin/CLAUDE.md` and `docs/admin-monorepo-migration.md`. (The former standalone `mx-space/mx-admin` repo is archived.) - **Frontend (Yohaku)**: `../Yohaku` — 主站前端 (Next.js) - **haklex**: `../haklex` (standalone) — Rich editor packages (`@haklex/*`) ### Lexical Content Processing mx-core uses
NVIDIAAGENTS.mdNVIDIA/TensorRT-Edge-LLM/AGENTS.md
tokenizer/`, `multimodal/`, `profiling/`, `sampler/`. **Python package (`tensorrt_edgellm/`)** is the checkpoint-based export frontend. It implements LLM architectures from scratch using ONNX builtin + custom ops (the only format EdgeLLM
nodetool-aiAGENTS.mdnodetool-ai/nodetool/AGENTS.md
NodeTool — Agent Rules Visual AI workflow platform. TypeScript monorepo with a React frontend, Electron desktop app, and Node.js backend. Read [Development Standards](docs/DEVELOPMENT_STANDARDS.md) for canonical engineering rules and targets, then
nodetool-aiAGENTS.mdnodetool-ai/nodetool/packages/AGENTS.md
watch, auto-restart) npm run dev:watch:server # Start both server and web frontend npm run dev:watch npm run dev # Start backend only npm run dev:watch:server
solo-agentAGENTS.mdsolo-agent/solo/AGENTS.md
complete architecture design covering the domain model, ownership and lifecycle, data flow, persistence, APIs, frontend state, failure recovery, migration, and compatibility with existing Solo behavior. - Review every change from
nirholasAGENTS.mdnirholas/XActions/AGENTS.md
Agent Skills, one directory each api/ Express backend (routes/, services/, middleware/) dashboard/ Static frontend scripts/ The 95 browser console scripts, plus build and maintenance scripts (twitter/ holds standalone console variants
CranotAGENTS.mdCranot/roam-code/AGENTS.md
Bridge auto-discovery + detection bridge_salesforce.py # Apex → Aura/LWC/Visualforce bridge bridge_protobuf.py # .proto → Go/Java/Python stubs bridge bridge_rest_api.py # Frontend HTTP calls → backend route definitions bridge_template.py # Jinja2/Django/ERB/Handlebars variable + include resolution bridge_django.py # Django admin/serializer/form/URL → Model + view
VikashLoombaAGENTS.mdVikashLoomba/copilot-mcp/AGENTS.md
clear validation over quick patches. ## Project Shape - VS Code extension backend: `src/` - React webview frontend: `web/src/` - Extension/webview RPC contracts: `src/shared/types/rpcTypes.ts` - Chat participant (`@mcp`): `src/McpAgent.ts` - Secondary Side Bar webview + Activity
agentregistry-devAGENTS.mdagentregistry-dev/agentregistry/AGENTS.md
servers, agents, and skills. **Tech Stack:** - **Backend/CLI:** Go 1.26+ - **Database:** PostgreSQL (accessed via pgx) - **Frontend:** Next.js 14 (App Router) with Tailwind CSS - **CLI Framework:** Cobra - **API Framework:** Huma (OpenAPI) --- ## Architecture
superdesigndevAGENTS.mdsuperdesigndev/superdesign-skill/AGENTS.md
path** (repo init is mandatory) and a **no-codebase path** (empty/scratch/sandbox workspace with no frontend code - skip init, gather design context conversationally, design via `SUPERDESIGN.md` "SOP: BRAND NEW PROJECT"). Before
cubeplexaiAGENTS.mdcubeplexai/cubeplex/AGENTS.md
agent platform. | Path | Role | |---|---| | `backend/` | FastAPI + cubeloop streaming runtime, SSE API, Postgres history, MCP | | `frontend/` | Next.js + React 19 monorepo (`packages/web`, `@cubeplex/core`) | | `docs/` | Cross-cutting docs; living process docs at `docs
dcoupleAGENTS.mddcouple/Pane/AGENTS.md
Project Structure & Module Organization - Root `pnpm` workspace with packages: `main/` (Electron main process, TypeScript), `frontend/` (React + Vite), `shared/` (shared types), and `tests/` (Playwright E2E). - Key paths: `main/src/{services,ipc,utils
managedcodeAGENTS.mdmanagedcode/dotnet-skills/AGENTS.md
browse flows should make it obvious whether the user is looking at `.NET`, frontend, testing, architecture, data, AI, or migration guidance. - Favor more explicit stacks over fewer overloaded ones
flytohubAGENTS.mdflytohub/flyto-core/AGENTS.md
handoffs; new handoffs use `YYYY-MM-DD-topic.md`. When changing behavior, public copy, deployment, security posture, or frontend UX, update the relevant memory files in the same change. Do not leave stale brand
r3bl-orgAGENTS.mdr3bl-org/r3bl-open-core/.agents/AGENTS.md
Module structure IS the API** - Different domains should be explicit: ```rust pub mod frontend; // Frontend-specific APIs pub mod backend; // Backend-specific APIs ``` 2. **Large feature domains** - When namespacing provides
An open format for instructions to coding agents, read by Codex, Cursor and others. Think of it as a README written for agents.
At the repository root, with more specific files in subdirectories. Agents read the one closest to the file they're editing.
Setup and test commands, code style, and the rules a new contributor would need to know.
Claude Code reads CLAUDE.md. A one-line CLAUDE.md that points at AGENTS.md covers both.
