agentleFS
xsovad06xsovad06/sova/templates/AGENTS.md

sova / templates

AGENTS.md3 starsChanged 4 days agoApache-2.0View on GitHub

Cross-AI-tool guidance. Works with Claude Code, Cursor, Copilot, or any AI coding assistant. Generated by SOVA (sova setup) {{PROJECT_DESCRIPTION}} {{TECH_STACK}} {{ARCHITECTURE_NOTES}} Remove rows for domains that don't apply. Add custom domains as needed. This project uses a four-tier knowledge management system. See KNOWLEDGE.md for full architecture. Promotion flow: Tier 2 (confirmed in 2+ tasks) → Tier 1 (confirmed in 2+ projects) → Tier 0. Each file < 150 lines. Confirmed, stable patterns only. Add more as the project grows. {{#TASK_SOURCE_GITHUB}}…

The file

Show its rules and text

We split this AGENTS.md into the rules it gives an agent

13 rules. About 0 are common, 8 are distinctive, and 5 are very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈8

    Distinctive · estimate

    This author's own choices. Worth a look.

  • 5

    Very specific

    Tied to their project's names, files or tools. Probably skip.

  • Common
  • Distinctive
  • Very specific
  • could be automatic: a setting could enforce it
  1. Very specific:

    Use conventional commits: type(scope): description

    line 92
  2. Very specific:

    Do NOT add "Co-Authored-By" lines for AI {{/NO_AI_COAUTHOR}} {{#AI_COAUTHOR}}

    line 99
  3. Very specific:

    Add "Co-Authored-By: Claude <noreply@anthropic.com>" to commits {{/AI_COAUTHOR}}

    line 102
  4. Very specific:

    No emoji characters in code or documentation {{/NO_EMOJIS}}

    line 117
  5. Distinctive:

    Follow existing code style and patterns

    line 119
  6. Distinctive:

    Keep functions focused and small

    line 120
  7. Distinctive:

    Don't add features beyond what the issue asks for

    line 122
  8. Distinctive:

    Always run tests before pushing

    line 125
  9. Very specific:

    Write tests for new features (TDD preferred) {{#COVERAGE_THRESHOLD}}

    line 126
  10. Distinctive:

    Don't modify code unrelated to the current task

    line 132
  11. Distinctive:

    Don't add dependencies without discussing first

    line 133
  12. Distinctive:

    Don't skip tests or lower coverage

    line 134
  13. Distinctive:

    Don't push without running the full test suite

    line 135

≈ “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.

Keep what fits and drop the rest. Your version, ready for your agents.
Files
AGENTS.md

{{PROJECT_NAME}} — Agent Instructions

Cross-AI-tool guidance. Works with Claude Code, Cursor, Copilot, or any AI coding assistant. Generated by SOVA (sova setup)

Project Overview

{{PROJECT_DESCRIPTION}}

Tech Stack

{{TECH_STACK}}

Architecture

{{ARCHITECTURE_NOTES}}

Key Paths

PathPurpose
{{KEY_PATHS}}

Domain Guidelines Index

FileScope
{{#DOMAIN_GUIDELINES}}
{{PATH}}{{SCOPE}}
{{/DOMAIN_GUIDELINES}}

Remove rows for domains that don't apply. Add custom domains as needed.

Context Index

FilePurpose
.claude/CLAUDE.mdClaude Code-specific commands and behaviors
.claude/commands/Workflow commands (develop, test, PR, review, etc.)
.claude/rules/*.mdModular knowledge files (architecture, patterns, testing)
CONTRIBUTING.mdContribution workflow, code style, PR format
README.mdProject overview, setup, getting started

Knowledge System

This project uses a four-tier knowledge management system. See KNOWLEDGE.md for full architecture.

TierLocationScopeIn Git?
0~/.claude/shared-knowledge/Cross-project patternsNo
1AGENTS.md, CLAUDE.md, .claude/rules/*.mdProject conventions (always loaded)Yes
2.claude/agent-memory/Agent learnings (loaded at task start)No
3~/.claude/projects/.../memory/User session memory (auto-managed)No

Promotion flow: Tier 2 (confirmed in 2+ tasks) → Tier 1 (confirmed in 2+ projects) → Tier 0.

Rules Files

FileScope
.claude/rules/architecture.mdApp structure, key paths, architectural decisions
.claude/rules/patterns.mdFramework/language gotchas and lessons learned
.claude/rules/testing.mdTest framework, fixtures, factories, conventions

Each file < 150 lines. Confirmed, stable patterns only. Add more as the project grows.

Project Tracker

{{#TASK_SOURCE_GITHUB}} This project uses GitHub Issues for task tracking.

  • Issue commands: gh issue list, gh issue view <NUMBER> {{/TASK_SOURCE_GITHUB}} {{#TASK_SOURCE_JIRA}} This project uses JIRA for task tracking. {{/TASK_SOURCE_JIRA}} {{#TASK_SOURCE_LINEAR}} This project uses Linear for task tracking. {{/TASK_SOURCE_LINEAR}} {{#TASK_SOURCE_MANUAL}} This project uses manual task input. {{/TASK_SOURCE_MANUAL}}

Conventions

Branches

  • Format: {{BRANCH_FORMAT}} (e.g., feat/42-add-user-auth)
  • Base branch: {{BASE_BRANCH}}

Commits

{{#CONVENTIONAL_COMMITS}}

  • Use conventional commits: type(scope): description
  • Types: feat, fix, refactor, test, docs, chore, style, perf, ci {{/CONVENTIONAL_COMMITS}} {{#FREEFORM_COMMITS}}
  • Descriptive commit messages, imperative mood {{/FREEFORM_COMMITS}} {{#NO_AI_COAUTHOR}}
  • Do NOT add "Co-Authored-By" lines for AI {{/NO_AI_COAUTHOR}} {{#AI_COAUTHOR}}
  • Add "Co-Authored-By: Claude noreply@anthropic.com" to commits {{/AI_COAUTHOR}}

PRs

  • Title: {{PR_TITLE_FORMAT}} {{#PR_AUTO_LINK}}
  • Include "Closes #N" in PR body to auto-link issues {{/PR_AUTO_LINK}}
  • Include a test plan section

Code Quality

{{#TYPE_HINTS}}

  • Type hints required on all new functions {{/TYPE_HINTS}} {{#NO_EMOJIS}}
  • No emoji characters in code or documentation {{/NO_EMOJIS}}
  • Follow existing code style and patterns
  • Keep functions focused and small
  • Make minimal, focused changes — don't refactor unrelated code
  • Don't add features beyond what the issue asks for

Testing

  • Always run tests before pushing
  • Write tests for new features (TDD preferred) {{#COVERAGE_THRESHOLD}}
  • Coverage threshold: {{COVERAGE_THRESHOLD}}% {{/COVERAGE_THRESHOLD}}

What NOT to Do

  • Don't modify code unrelated to the current task
  • Don't add dependencies without discussing first
  • Don't skip tests or lower coverage
  • Don't push without running the full test suite

Important Patterns

{{PATTERNS}}

Invariants

These constraints are checked automatically before push: {{INVARIANT_LIST}}

Persona

{{PERSONA_GUIDANCE}}

Agentic Workflow Commands

This repo has Claude Code commands in .claude/commands/:

Core Development

  • /develop — implement a feature or fix (TDD approach)
  • /develop-full — full workflow: develop + test + review + PR
  • /develop-explain — explain approaches without implementing
  • /test — run and fix tests iteratively
  • /review — pre-push code review with auto-fix
  • /debug — systematic debugging workflow

Pull Requests

  • /pr — create PR with standard template
  • /address-pr — address PR review comments (human and bot)
  • /review-pr — review a teammate's PR

Project Management

  • /standup — daily context summary
  • /find-task — browse issue backlog and prioritize sprint work
  • /issue — fetch and analyze a GitHub Issue
  • /new-feature — set up a new feature branch
  • /status — project health check

Post-Work

  • /after-merge — post-merge cleanup
  • /rearrange-commits — reorganize branch commits
  • /ingest-review — learn from PR feedback
  • /extract-knowledge — capture reusable patterns

Meta

  • /agent-readiness — assess and improve AI readiness

Download now (ZIP)

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
What did you use it for? e.g. Reviewing pull requests in a Rails app

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.