harness-starter-kit
Harness Starter Kit is a prompt-first toolkit that helps developers turn repeated coding-agent mistakes into durable repository instructions, checks, memory, and evaluation. It provides workflows and commands to safely isolate agent work, diagnose failures, and create an improvement loop. It is for developers using coding agents like Codex or Claude Code who want to make agent work safer and more effective.
✨ Key features
- Prompt-first harness engineering workflow for repositories.
- Commands: doctor, adopt, review, update, refresh.
- Runtime-native skills for Codex and Claude Code.
- Templates for profiles, checklists, and effectiveness reports.
- Optional Python installer for skeleton profiles.
- Multi-language README support.
🎯 Use cases
- Adopt harness practices in a repository to guide coding agents.
- Diagnose agent failures with structured failure types.
- Review diffs before commit or PR from an opposing perspective.
- Update or refresh harness guidance as the kit evolves.
- Measure agent effectiveness with task outcomes and reports.
📦 Installation
🧰 Requirements: Requires a coding agent (e.g., Codex or Claude Code) and Python for the optional installer. No API keys or accounts stated.
Codex
codex plugin marketplace add harnessworks/harness-agent-skills-marketplace --ref v0.1.16
Restart Codex, open the Plugins screen, select Harnessworks, and install harness-agent-skills.
Claude Code
claude plugin marketplace add harnessworks/harness-agent-skills-marketplace@v0.1.16
claude plugin install harness-agent-skills@harnessworks
Optional Installer
python harness-starter-kit/scripts/apply_harness.py --target . --profile generic --dry-run
🚀 Usage
Open the target repository with your coding agent and give it this prompt:
Use this kit to apply harness engineering to this repository:
https://github.com/harnessworks/harness-starter-kit
Clone the kit into ./harness-starter-kit if it is not already present, read it,
then apply its prompt-first harness engineering workflow to this repository.
Requirements:
- Treat the current working directory as the target repository.
- Treat ./harness-starter-kit as read-only reference material after cloning.
- Inspect this repository before editing.
- Preserve existing architecture, tools, package manager, commands, docs, and
conventions.
- Do not blindly copy templates.
- Add only the minimum useful harness pieces.
- Prefer updating existing docs/configs over duplicating them.
- Do not overwrite or delete existing files without explaining why.
- If I ask for /harness doctor, use
./harness-starter-kit/commands/harness-doctor.md.
- If I ask for /harness update after adoption, use
./harness-starter-kit/commands/harness-update.md to refresh the kit reference,
record .harness/source.json, and selectively update target harness files
without blindly overwriting existing files.
-
⚠️ Good to know
The kit is prompt-first and not an automatic installer; the optional installer only copies profile snippets for review. Validation and lifecycle pilot results do not prove that harness adoption reduces repeated agent mistakes.
❓ FAQ
What is the difference between /harness doctor and /harness adopt?
Doctor inspects current harness readiness without modifying files, while adopt applies the smallest useful harness pieces to the repository.
Are the /harness commands built-in editor commands?
No, they are prompt conventions by default. In editors like Cursor, they won't appear in the command palette unless you add matching custom slash commands.
How do I install the agent skills for Codex?
Run codex plugin marketplace add harnessworks/harness-agent-skills-marketplace --ref v0.1.16, restart Codex, open the Plugins screen, select Harnessworks, and install harness-agent-skills.
What should I do when the kit version changes?
Use the /harness update command to bring in a newer harness-starter-kit reference and selectively update the target repository's harness files.
📊 Repository
🤖 Overview, features, install steps and FAQ were generated from the project's README on Sep 4, 2026. Always check the original source before running commands.