← Browse

@ganimjeong/harness-for-codex

A

Codex Harness Instructions

instructionscodexclaude

Install

agr install @ganimjeong/harness-for-codex --target claude

Writes 1 file into .claude/skills/, pinned to git-839a55b9.

  • .claude/skills/harness-for-codex/AGENTS.md

Document

Codex Harness Instructions

This repository is a baseline workspace for Codex-driven implementation.

Commands

Run these from the repository root:

scripts/bootstrap
scripts/check
scripts/test
scripts/eval
scripts/doctor
scripts/hooks
  • scripts/bootstrap: prepare dependencies when a known stack is present.
  • scripts/check: run lint, type checks, formatting checks, and tests when available.
  • scripts/test: run only the test suite when available.
  • scripts/eval: run the complete handoff verification sequence.
  • scripts/doctor: report repository, tool, and environment readiness.
  • scripts/hooks: install local Git hooks through pre-commit when available.

Repository Layout

  • AGENTS.md: Codex instructions for this repo.
  • CLAUDE.md: Claude Code bridge that imports AGENTS.md.
  • README.md: Human-facing overview and workflow.
  • harness.yml: Machine-readable harness metadata and command registry.
  • .devcontainer/: Optional reproducible development container.
  • .pre-commit-config.yaml: Optional local hook definitions.
  • docs/: Project notes and decisions.
  • scripts/: Stable automation entrypoints.
  • tasks/: Task briefs and working notes.

Operating Principles

  • Read the repository before editing. Prefer existing conventions over new ones.
  • Keep changes scoped to the requested task.
  • Do not revert user changes unless explicitly asked.
  • Use rg for searching when available.
  • Use the standard scripts instead of one-off local commands when they cover the task.
  • Add or update tests when behavior changes.
  • Document decisions that affect future work in docs/decisions.md.
  • Add nested AGENTS.md files in subdirectories only when that area needs different instructions.
  • Keep root-level agent instructions short and durable. Move long procedures into docs or scripts.

Task Loop

  1. Inspect: read relevant files and recent decisions.
  2. Plan: identify the narrow change and verification path.
  3. Implement: make focused edits.
  4. Verify: run scripts/check or scripts/eval.
  5. Handoff: summarize changed files, verification, and residual risks.

Cross-Agent Compatibility

  • Codex and Cursor can read AGENTS.md.
  • Claude Code reads CLAUDE.md, which imports AGENTS.md.
  • Keep shared instructions in AGENTS.md unless a tool-specific note is required.

Adopting This Harness

After creating a repository from this template or fork:

  1. Replace the LICENSE copyright line with the new project owner's name.
  2. Rewrite README.md for the new project; the default README describes the harness itself.
  3. Update repository metadata such as description, topics, and social preview.
  4. Remove or adjust .github/ISSUE_TEMPLATE entries that do not fit the project.
  5. Keep the standard scripts unless the new project has a better documented entrypoint.

Script Policy

These scripts are intentionally stack-aware and conservative. If a language stack is added later, extend the scripts rather than bypassing them.

Completion Criteria

Before finishing a task:

  1. Confirm the requested change is implemented.
  2. Run the narrowest useful verification command, usually scripts/check.
  3. Report changed files and any verification that could not be run.

Trustgrade A

  • passBody integrity

    Whether the stored document is plausibly the kind of file the artifact declares, rather than something fetched by mistake.

  • passType matchnot applicable to this artifact type

    Whether the artifact is really the kind of thing its metadata claims it is.

  • passFreshness

    How long since the source repository was last pushed to.

  • passPrompt injection

    Scans the artifact's own text for instructions aimed at your agent rather than at you.

  • passLicense

    Whether the source repository declares an SPDX license permissive enough to redistribute.

How the grade is calculated

Each check contributes 0 points when it passes, 1 when it warns, and 2 when it fails. The total maps to a letter:

  • Aevery check passed
  • Bone warning
  • Ctwo warnings
  • Dprompt injection or body integrity failed, or three warnings
  • Fone of those failed, and something else is wrong

These are automated hygiene checks, not a security audit, and not a dependency or vulnerability scan. A grade of A means nothing was flagged — not that the artifact is safe.

Versions

  • git-839a55b9971d2026-08-04