An Agent Skill that gives AI coding agents a spec-driven project workflow, turning PRDs into epics and GitHub Issues, with parallel execution in Git worktrees.
8.4K+11 this week841ShellMITUpdated 3 days ago
ericrisco/rsc-harness
A CLI that installs a project-local harness of skills, rules, hooks and knowledge docs for 18 AI coding assistants, after you review the plan.
- Stars
- 521
- Forks
- 73
- Open issues
- 5
- Last commit
- 3 days ago
What is it?
rsc-harness is a meta-harness: it sets up the instructions, skills, working rules and project documentation that AI coding assistants rely on, tailored to your project. It asks a few onboarding questions about your goal and assistants, then shows a plan in which each component is selected, deferred or excluded with a reason. Nothing is written until you accept that exact plan.
The setup lives inside the project. Skill files are stored once under .rsc/skills and linked into each assistant's own folder, so choosing several assistants does not duplicate content. The catalog includes 274 skills covering languages, frameworks, databases, deployment, testing, and business tasks such as invoicing and sales. A local 02-DOCS wiki holds project knowledge, and a 01-TOOLS convention documents how external services are connected.
In Claude Code, hooks enforce some rules, such as refusing commits to a closed main branch and confirming destructive commands. In other assistants, the same rules are followed through skills and instructions but not enforced by hooks. Session memory and team knowledge sync are supported for several targets, with differing levels of support listed per assistant.
Key features
- 01Installs project-local skills, rules and hooks for 18 AI assistant targets.
- 02Shows a reviewable plan with selected, deferred and excluded components before writing files.
- 03Stores each skill once and links it into every assistant's folder.
- 04Provides a catalog of 274 skills spanning development, operations and business tasks.
- 05Keeps project knowledge in a local markdown wiki that agents consult before acting.
- 06Offers a doctor command to check harness health and repair drift.
Use cases
- Setting up a consistent set of agent instructions and skills across a team using several AI coding assistants.
- Giving an agent a maintained project knowledge base and documented connections to external tools.
- Adding guardrails such as protected-branch rules and confirmation before destructive commands in Claude Code.
- Adding individual skills to an existing project without adopting the full workflow.
In collections
Topics
FAQ
- What is rsc-harness?
- A CLI that installs a project-local harness of skills, rules, hooks and knowledge docs for 18 AI coding assistants, after you review the plan. rsc-harness is a meta-harness: it sets up the instructions, skills, working rules and project documentation that AI coding assistants rely on, tailored to your project. It asks a few onboarding questions about your goal and assistants, then shows a plan in which each component is selected, deferred or excluded with a reason. Nothing is written until you accept that exact plan.
- Is rsc-harness free to use? What license does it have?
- rsc-harness is open source under the MIT license. Check the license terms before using it in commercial products.
- Is rsc-harness still maintained?
- Actively maintained. The last commit to rsc-harness was on Nov 4, 2026.
- What can I use rsc-harness for?
- Setting up a consistent set of agent instructions and skills across a team using several AI coding assistants. Giving an agent a maintained project knowledge base and documented connections to external tools. Adding guardrails such as protected-branch rules and confirmation before destructive commands in Claude Code. Adding individual skills to an existing project without adopting the full workflow.
- Where can I get rsc-harness?
- The source code and installation instructions are on GitHub: https://github.com/ericrisco/rsc-harness
Summary generated by AI from the project's README. GitHub stats refresh daily.
Related projects
A CLI app for running and switching between multiple AI coding agent sessions across Git worktrees and projects, without tmux.
1.3K+2 this week94TypeScriptMITWell maintainedUpdated 11 days ago
A set of agent skills that take an idea or existing codebase through a PRD, features, rules, and sequenced RFCs before AI coding agents write code.
309+3 this week33PythonMITUpdated 30 days ago
An agentic IDE built around the pi coding agent, running each task in isolated git-worktree workspaces with living specs and reusable skills.
7.8K+1.4K this week478TypeScriptApache-2.0Updated 21 hours ago
A pattern library and CLI toolkit for designing loops that let AI coding agents discover work, act on it, verify results and persist state.
11.7K+41 this week1.6KTypeScriptMITWell maintainedUpdated yesterday
Installs a spec-driven development workflow as Agent Skills for AI coding agents like Claude Code, Codex, and Cursor, covering requirements, design, tasks, and autonomous implementation.
3.8K+7 this week286TypeScriptMITUpdated 16 days ago