Community
Explore custom schemas and projects built by the OpenSpec community.
- Custom schema examples: workflows you can adapt to your project.
- Community projects: open-source tools and extensions.
These projects are maintained independently. A listing does not imply OpenSpec support or endorsement. Use each project's documentation and issue tracker for setup and help.
Custom schema examples
These schemas live in their maintainers' repositories and are released separately from OpenSpec. Follow the repository's installation and compatibility notes. For how schema folders work, see Schemas.
Planning and coordination
- intent-driven by @harikrishnan83: Records change intent, observable behavior, technical design, and architectural decisions. Includes a review manifest for decisions that should become durable ADRs.
- nanopm by @nmrtn: Connects product planning to implementation. Uses nanopm's audit, strategy, roadmap, and PRD as inputs to OpenSpec artifacts.
- epic-driven by @ivanferic: Splits an epic into child stories. Plan, implement, and archive the stories individually, then sync their progress back to the epic. The epic has no apply phase.
Review and execution
- superpowers-bridge by @JiangWay: Connects OpenSpec artifacts to Superpowers execution skills, including planning, TDD, and code review. Adds a retrospective artifact for recording evidence from the work.
- anvil by @jikkujoyce: Adds an adversarial review and a test plan before implementation. Maps spec scenarios to tests and tracks red/green results for verification.
- egm by @lutfi-zain: Adapts Google's Elephant-Goldfish Model. Instructs fresh-context reviewers to run five review stages and reach two clean rounds before the final review. Its flow is proposal → specs and design → review → tasks → apply.
Review gates: anvil and egm give the agent review instructions. OpenSpec checks that artifacts exist, not that a reviewer approved them. Use your own CI or hook if approval must be enforced.
Testing and research
- e2e-runbooks by @Lukk17: Defines end-to-end test runbooks for each capability. Keeps a spec, a task template, and a timestamped record of each execution, with assertions about observable behavior.
- ml-research by @tsenst: Plans machine-learning experiments. Sets evaluation criteria before the run and records evidence-backed findings afterward, including refuted hypotheses.
Community projects
Tools and integrations
- OpenSpec Workbench: A UI for running and supervising agents working on OpenSpec changes.
- openspec-guard: A CLI and GitHub Action that reports which spec scenarios have Vitest or Jest coverage. It does not run the tests.
- rightcommit: A self-hosted, Apache-2.0 orchestrator with a guide to running OpenSpec's propose, apply, and archive skills through Claude Code, Codex, or Pi. An optional human review step sits before implementation. Requires a local account and your own agent API keys or authorized provider access.
Add a schema or project
Open an issue with the name, a public source repository link, and a short description of how it supports OpenSpec. For a schema, include its maintainer and what the workflow changes. Maintainers review and add entries here.
- Describe what people can use. Leave out promotional claims, referral links, and tracking links.
- Include an open-source license and a documented OpenSpec integration or example.
- The OpenSpec functionality must work without requiring the project's proprietary hosted service.
- Disclose any required coding-agent accounts or credentials.
Use the same issue process to correct or update a listing.