agentleFS
Sign inStart free
tardigrdetardigrde/agent-skill-eval/AGENTS.md

agent-skill-eval

AGENTS.md4 starsChanged 5 months agoMITView on GitHub

Python CLI framework (agent-skill-eval) for evaluating agent skills across OpenCode, Claude Code, and Codex. Uses Typer, Pydantic, OpenAI SDK, and Rich.

  • Reads credentials
  • Installs packages
  • 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 31,774 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 repository5 other files it gives its agents.

The file

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

3 rules. About 0 are common, 0 are distinctive, and 3 are 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.

  • 3

    Very specific

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

  • Common
  • Distinctive
  • Very specific
  1. Very specific:

    Test files mirror source: test_runner.py, test_graders.py, test_harnesses.py, etc.

    line 44
  2. Very specific:

    All imports use from __future__ import annotations

    line 54
  3. Very specific:

    Default grader model deepseek/deepseek-v4-flash is a valid OpenRouter model.

    line 61

≈ “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

AGENTS.md

Project

Python CLI framework (agent-skill-eval) for evaluating agent skills across OpenCode, Claude Code, and Codex. Uses Typer, Pydantic, OpenAI SDK, and Rich.

Quick commands

# Install (editable + dev deps)
uv pip install -e ".[dev]"

# Run tests
uv run --extra dev pytest -q

# Lint + format check (CI runs both)
uv run --extra dev ruff check src/ tests/
uv run --extra dev ruff format --check src/ tests/

# Run all checks before committing
uv run --extra dev ruff check src/ tests/ && uv run --extra dev ruff format --check src/ tests/ && uv run --extra dev pytest -q

Architecture

src/agent_skill_eval/
├── cli.py            # Typer CLI app (run, report, grade, cleanup, init)
├── runner.py         # EvalRunner orchestrates eval execution
├── graders/__init__.py   # DeterministicGrader + LLMGrader (700+ lines, single file)
├── harnesses/__init__.py # Agent harnesses (opencode, claude-code, codex, fake)
├── models.py         # Pydantic models for eval config
├── git_state.py      # Pre/post git/PR state snapshots
├── skills.py         # Skill installer
└── workspace.py      # Workspace creation (fresh git init or clone)
  • Entry point: agent_skill_eval.cli:app (registered as agent-skill-eval console script)
  • Package layout: src/agent_skill_eval/ (hatchling build, wheel packages src/agent_skill_eval)
  • The fake harness (used in tests) is a no-op agent that produces deterministic output

Testing

  • Test files mirror source: test_runner.py, test_graders.py, test_harnesses.py, etc.
  • Fixtures live in tests/fixtures/ — skills and evals JSON used by smoke tests
  • test_cli_smoke.py is the integration test: runs full run -> report -> init cycle with --agent fake
  • Tests use tmp_path for workspace isolation; no shared state
  • Shared fixtures in tests/conftest.py (_init_git_workspace, grader, evals_path)

Conventions

  • Python 3.11+ required (requires-python = ">=3.11")
  • Ruff config: line-length = 120, rules ["E", "F", "I"]
  • All imports use from __future__ import annotations
  • Pydantic v2 models throughout
  • CLI uses Typer with CliRunner for testing
  • .env file loaded via python-dotenv (not committed)

Key gotchas

  • Default grader model deepseek/deepseek-v4-flash is a valid OpenRouter model. Override with --grader-model or set a real model.
  • graders/__init__.py is a monolith (700+ lines) containing both deterministic and LLM graders
  • harnesses/__init__.py has a dead cost_usd read at line ~140 (no-op code)
  • compute_benchmark returns delta=0.0 for multi-agent runs (known bug)
  • Workspace paths must contain literal with_skill/without_skill substrings or grading breaks silently
  • uv.lock is gitignored — not tracked in version control

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.

Check your setup, free