agentleFS
thiennc-tesoglobalthiennc-tesoglobal/flutter-skills/AGENTS.md

flutter-skills

AGENTS.md16 starsChanged 2 months agoBSD-3-ClauseView on GitHub

This repository contains Agent Skills for Flutter and Dart app delivery. It is not a Flutter application or reusable Dart package. Treat changes as instruction, routing, evaluation, and distribution work. Before creating or substantially revising a skill: When implementing any feature, bug fix, refactor, evaluation, or documentation update: When adding, removing, or renaming a skill, updat

  • Commits and pushes

1 other project uses a version of this.

Adopted by

Most recently taken up by

The file

Show its rules and text

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

23 rules. About 1 is common, 16 are distinctive, and 6 are very specific to this project.

  • ≈1

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈16

    Distinctive · estimate

    This author's own choices. Worth a look.

  • 6

    Very specific

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

  • Common
  • Distinctive
  • Very specific
  1. Distinctive:

    Before creating or substantially revising a skill:

    line 9
  2. Distinctive:

    Follow the active Agent Skills authoring guidance available in the agent environment.

    line 11
  3. Distinctive:

    Verify fast-moving Flutter, Dart, package, and platform claims against current primary sources.

    line 13
  4. Distinctive:

    Add or update behavior-focused evaluation cases.

    line 15
  5. Distinctive:

    When implementing any feature, bug fix, refactor, evaluation, or documentation update:

    line 19
  6. Distinctive:

    Before starting, ensure main is up to date:

    line 22
  7. Distinctive:

    Create and checkout a descriptive topic branch with an appropriate prefix:

    line 26
  8. Distinctive:

    Run the required validation commands:

    line 39
  9. Very specific:

    Use conventional commit messages: feat(...), fix(...), docs(...), test(...), refactor(...).

    line 49
  10. Distinctive:

    Create a Pull Request with a clear summary and verification plan:

    line 56
  11. Distinctive:

    Only merge after all CI checks pass:

    line 66
  12. Very specific:

    Keep SKILL.md concise and place conditional detail in directly linked references.

    line 77
  13. Distinctive:

    Do not promote Provider, Riverpod, Bloc, Dio, Drift, go_router, or another third-party package as a universal default.

    line 80
  14. Very specific:

    Read pubspec.yaml and SDK constraints before recommending syntax or packages.

    line 81
  15. Distinctive:

    Prefer current official Flutter and Dart documentation.

    line 82
  16. Very specific:

    Ensure all reference documents in references/ are covered in .github/evals/behavior/<name>/cases.json (python3 …

    line 84
  17. Very specific:

    Keep maintainer-only workflows outside skills/ so they do not enter the public catalog.

    line 86
  18. Distinctive:

    When adding, removing, or renaming a skill, update:

    line 90
  19. Distinctive:

    When forward-evaluation behavior changes, run the smallest relevant executed case or the fixed public profile and retain the raw result.

    line 116
  20. Very specific:

    Before publishing, replace the version's Unreleased marker with an ISO date and run python3 .github/scripts/validate_repository.py …

    line 120
  21. Distinctive:

    Do not publish packages, create releases, change signing material, upload apps, or mutate live services without explicit authorization.

    line 125
  22. Distinctive:

    Do not claim device, build, test, or performance verification that was not actually performed.

    line 126
  23. Common:

    If credentials or platform tooling block verification, state the exact boundary instead of weakening the check.

    line 127

≈ “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

Repository instructions

Purpose

This repository contains Agent Skills for Flutter and Dart app delivery. It is not a Flutter application or reusable Dart package. Treat changes as instruction, routing, evaluation, and distribution work.

Required authoring workflow

Before creating or substantially revising a skill:

  1. Follow the active Agent Skills authoring guidance available in the agent environment.
  2. Inspect the affected skill, neighboring descriptions, evaluation cases, package manifests, and working tree.
  3. Verify fast-moving Flutter, Dart, package, and platform claims against current primary sources.
  4. Preserve existing project architecture and package choices unless the user's task explicitly requests migration.
  5. Add or update behavior-focused evaluation cases.

Git, branching, and pull request workflow

When implementing any feature, bug fix, refactor, evaluation, or documentation update:

  1. Always work in a separate branch (Never commit directly to main):
    • Before starting, ensure main is up to date:
      git checkout main && git pull origin main
    • Create and checkout a descriptive topic branch with an appropriate prefix:
      • feature/<name> for new capabilities, skills, or tooling (e.g. feature/eval-coverage-gate)
      • fix/<name> for bug fixes, corrections, or broken links (e.g. fix/device-cleanup-guidance)
      • refactor/<name> for restructuring without functional changes
      • docs/<name> for documentation-only updates
      • Or <username>/<topic> (e.g. thiennc/phase2-eval-engine)
      git checkout -b feature/<feature-name>
      # or
      git checkout -b fix/<fix-name>
  2. Validate locally before committing:
    • Run the required validation commands:
      python3 .github/scripts/validate_repository.py
      python3 .github/scripts/run_behavior_evals.py --coverage
      python3 -m unittest discover -s tests -v
      npm test
      npm run pack:check
  3. Commit with semantic and atomic messages:
    • Use conventional commit messages: feat(...), fix(...), docs(...), test(...), refactor(...).
  4. Push and create a Pull Request:
    • Push the branch to origin:
      git push -u origin <branch-name>
    • Create a Pull Request with a clear summary and verification plan:
      gh pr create --title "..." --body "..."
  5. Verify CI and merge:
    • Monitor GitHub Actions status:
      gh pr checks <pr-number> --watch
    • Only merge after all CI checks pass:
      gh pr merge <pr-number> --squash --delete-branch
    • Switch back to main and pull the latest merge commit:
      git checkout main && git pull origin main

Skill rules

  • Keep SKILL.md concise and place conditional detail in directly linked references.
  • Make frontmatter descriptions cheap, precise, and discriminating.
  • Define boundaries between nearby specialists to prevent broad activation.
  • Do not promote Provider, Riverpod, Bloc, Dio, Drift, go_router, or another third-party package as a universal default.
  • Read pubspec.yaml and SDK constraints before recommending syntax or packages.
  • Prefer current official Flutter and Dart documentation; use package-publisher documentation for package-specific behavior.
  • Require evidence proportionate to the claim: formatting, analysis, tests, profile-mode measurements, or device behavior.
  • Ensure all reference documents in references/ are covered in .github/evals/behavior/<name>/cases.json (python3 .github/scripts/run_behavior_evals.py --coverage must report 100%).
  • Designate critical correctness, security, or constraint expectations with mandatory: true to enforce strict evaluation gating.
  • Keep maintainer-only workflows outside skills/ so they do not enter the public catalog.

Surfaces that must stay synchronized

When adding, removing, or renaming a skill, update:

  • skills/<name>/SKILL.md
  • .github/evals/behavior/<name>/cases.json
  • .codex-plugin/plugin.json
  • .agents/plugins/marketplace.json
  • .tessl-plugin/plugin.json
  • README.md counts and catalog
  • CHANGELOG.md

Validation

Run before committing:

python3 .github/scripts/validate_repository.py
python3 .github/scripts/run_behavior_evals.py --coverage
python3 -m unittest discover -s tests -v
npm ci
npm test
npm run pack:check
bash .github/scripts/run_dart_skills_lint.sh
claude plugin validate .
npx skills add . --list

When forward-evaluation behavior changes, run the smallest relevant executed case or the fixed public profile and retain the raw result. Do not rewrite skills without a measured failure or a current primary-source change.

For a full audit or release review, follow the maintainer audit workflow.

Before publishing, replace the version's Unreleased marker with an ISO date and run python3 .github/scripts/validate_repository.py --release.

Safety and scope

  • Preserve unrelated user changes.
  • Do not publish packages, create releases, change signing material, upload apps, or mutate live services without explicit authorization.
  • Do not claim device, build, test, or performance verification that was not actually performed.
  • If credentials or platform tooling block verification, state the exact boundary instead of weakening the check.

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.