awesome-agentic-AI-coding-template
AGENTS.md4 starsChanged 12 days agoMITView on GitHub
Agent instructions for AI coding tools (OpenAI Codex, GitHub Copilot, Cursor, Kilo Code, Google Jules). For Claude Code see CLAUDE.md. For Kilo Code project rules see .kilocode/rules/. For Gemini/Antigravity see .agent/rules/. [FILL: Project Name] — [FILL: 1-2 sentence project description. What does it do? Who uses it?] [FILL: e.g. - api-variant-id-1 — used in deeplinks - SCHEMA_VERSION=3 — written to stored data ] Agent Skills (via /skill-name in chat) and Prompt Files (via /command) are available for common workflows:
Where files are headingNew AGENTS.md files have more rules about keeping secrets out of the code, and fewer about code style.
Of the AGENTS.md files we read, the 31,700 first seen in the last six months, against the older ones: the kind of rule that gained most share, and the kind that lost most. About files like this in general, not this one, and an estimate, since a kind is a pattern in the words.
In this repository34 other files it gives its agents.
- CLAUDE.mdCLAUDE.md
- .github/copilot-instructions.mdCopilot instructions
- architecture-changeSkill
- brainstormingSkill
- doc-auditSkill
- feature-deliverySkill
- find-skillsSkill
- frontend-designSkill
- hotfixSkill
- jules-tasksSkill
- qa-auditSkill
- release-readinessSkill
- requesting-code-reviewSkill
- setup-repoSkill
- skill-creatorSkill
- systematic-debuggingSkill
- using-superpowersSkill
- writing-plansSkill
- architecture-changeSkill
- brainstormingSkill
- doc-auditSkill
- feature-deliverySkill
- find-skillsSkill
- frontend-designSkill
- hotfixSkill
- jules-tasksSkill
- qa-auditSkill
- release-readinessSkill
- requesting-code-reviewSkill
- setup-repoSkill
- skill-creatorSkill
- systematic-debuggingSkill
- using-superpowersSkill
- writing-plansSkill
The file
We split this AGENTS.md into the rules it gives an agent
1 rule. About 0 are common, 0 are distinctive, and 1 is very specific to this project.
≈0
Common · estimate
Many files like this have a rule of this kind.
≈0
Distinctive · estimate
This author's own choices. Worth a look.
1
Very specific
Tied to their project's names, files or tools. Probably skip.
- Common
- Distinctive
- Very specific
- Very specific: line 94
Commit via git commit (hooks run automatically) with format type(scope): description
≈ “Common” is an estimate: at least 1 in 5 of the 40,246 AGENTS.md files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.
Files
AGENTS.md
Agent instructions for AI coding tools (OpenAI Codex, GitHub Copilot, Cursor, Kilo Code, Google Jules). For Claude Code see
CLAUDE.md. For Kilo Code project rules see.kilocode/rules/. For Gemini/Antigravity see.agent/rules/.
<!-- SETUP: Fill in all [FILL: ...] placeholders. See README.md for the full setup guide. -->Project Overview
[FILL: Project Name] — [FILL: 1-2 sentence project description. What does it do? Who uses it?]
Architecture (Must-Know)
<!-- Keep only what agents cannot infer from the code. Remove or replace examples below. -->- [FILL: Key architectural constraint 1] — e.g. "No runtime frameworks, vanilla JS only"
- [FILL: Key architectural constraint 2] — e.g. "Config-driven: all data in config/ files, never hardcode in templates"
- [FILL: Key architectural constraint 3] — e.g. "Touch-first: runs in iOS/Android WebView, all interactions are touch gestures"
- Repo-local Git hooks — Use
.githooks/viagit config core.hooksPath .githooksand keep hooks executable
File Structure
[FILL: Add your project's key file structure here.
Only include files/folders that agents must know about to work effectively.
Example:
src/
index.ts ← Entry point
api/ ← API layer
components/ ← UI components
config/
app-config.ts ← Application configuration
tests/
*.test.ts ← Unit tests
docs/
README.md ← Primary documentation
]Coding Rules
General
- [FILL: Rule 1] — e.g. "No runtime frameworks"
- [FILL: Rule 2] — e.g. "Config as Single Source of Truth"
- Minimal changes — Only touch files necessary for the task
- No unsolicited features — Don't add functionality the user didn't ask for
AI Execution & Orchestration
- No hanging bash loops — Never use fire-and-forget bash scripts for multi-step iteration or orchestration.
- Agent-Orchestrated Concurrency — For batch tasks, retrieve the list of targets via a native tool, then have the agent iterate with individual tool calls.
- MCP Server Tools Over CLI — Always prioritize native MCP integrations (e.g., Github MCP Server) over arbitrary CLI commands.
- Strict Subprocess Timeouts — Every CLI command via
child_processmust have a strict timeout (e.g.,{ timeout: 10000 }).
Documentation Hygiene
- Table Style — Use
| --- | --- |separator style for all tables. - Fenced Code Language — Every fenced code block must include a language specifier.
- Heading Quality — Single H1 per file (MD025), logical increments (MD001), unique subheadings within document (MD024).
Immutable Contract — NEVER CHANGE
<!-- Document any IDs, keys, or values that are used externally and must never change. --><!-- If none: remove this section. -->[FILL: e.g.
api-variant-id-1— used in deeplinksSCHEMA_VERSION=3— written to stored data ]
Common Tasks
Agent Skills (via /skill-name in chat) and Prompt Files (via /command) are available for common workflows:
| Task | Invoke | Details |
| Setup repo (AI-powered) | /setup-repo skill | .agents/skills/setup-repo/SKILL.md |
| QA audit | /qa-audit skill | .agents/skills/qa-audit/SKILL.md |
| Documentation audit | /doc-audit skill | .agents/skills/doc-audit/SKILL.md |
| Feature delivery | /feature-delivery skill | .agents/skills/feature-delivery/SKILL.md |
| Plan a new feature | /new-feature prompt | .github/prompts/new-feature.prompt.md |
| Review changes pre-commit | /review-changes prompt | .github/prompts/review-changes.prompt.md |
| Release readiness | /release-readiness skill | .agents/skills/release-readiness/SKILL.md |
| Hotfix | /hotfix skill | .agents/skills/hotfix/SKILL.md |
After Every Change
- Run
[FILL: your test command, e.g. npm test] - Commit via
git commit(hooks run automatically) with formattype(scope): description - [FILL: Any additional verification steps specific to your project]
Documentation
| Doc | What it covers |
| README.md | Project overview and quick start |
| docs/ARCHITECTURE.md | Template architecture and design decisions |
| docs/CUSTOMIZATION-GUIDE.md | How to fill every [FILL:] marker |
| docs/SKILLS-CATALOG.md | All available skills and how to invoke them |
| CHANGELOG.md | Version history |
| [FILL: Add links to project-specific docs, e.g. ENGINEERING-PLAYBOOK.md, API-REFERENCE.md] |
Does it work?
Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.
No reviews yet
People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.
- It worked
- Partly
- It didn't
Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.
Your trail
Files you open will collect here, so you can find them again.
Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.
Now do this for your own.
Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.

