Spec Kit Skills for Coding Runtimes
Bring Spec Kit’s specification-driven development workflow to several coding-agent runtimes.
- Source repo
- dceoy/speckit-agent-skills
- Stars
- ★ 151
- Last updated
- 1d ago
- License
- AGPL-3.0
- Primary language
- Shell
- FA score
- 51/100 · Major gaps
At a glance
- How it runs
- Works with
- Portable with changesCodex · Claude Code
- Setup effort
- Medium · a few setup steps
- You'll need
- Typical use
- Developers using Claude Code or Codex CLI can reuse the shared skills to follow a Spec Kit development workflow.
- Not a fit if
- Teams that need a native package for only one runtime
- Users who want to use the skills without installing Spec Kit and a runtime
- Source review
- 51/100 · Major gaps
What does this agent do, and when should you use it?
This repository tracks generated Spec Kit output alongside shared speckit-* skills and repository-specific skills in skills/. The workflow covers project principles, requirements, planning, tasks, and implementation, with optional clarification, analysis, convergence, checklists, and task-to-GitHub-Issues utilities. Claude Code and Codex CLI expose the shared skills through directory links; Cursor Agent, GitHub Copilot CLI, and OpenCode use runtime-specific generated skills, agents, prompt wrappers, or commands. Gemini CLI commands are checked in, but the current CI configuration does not regenerate them. Users install Spec Kit and initialize the target project for the chosen integration; this repository supplies skills and runtime entry points rather than a hosted service.
The workflow uses speckit-constitution to define project principles and speckit-specify to capture requirements for new work, or speckit-baseline to derive a specification from an existing codebase. Run speckit-clarify when material ambiguities need resolution, then speckit-plan to produce an implementation strategy, speckit-tasks to generate ordered work items, and speckit-implement to execute them. Optional utilities include speckit-analyze for consistency checks across the specification, plan, and tasks; speckit-converge to compare an implementation with those artifacts and append remaining work to tasks.md; speckit-checklist for requirements-quality checks; and speckit-taskstoissues to convert tasks into GitHub Issues. CI checks the latest Spec Kit version and regenerates configured integration outputs when that version changes.
- Developers using Claude Code or Codex CLI can reuse the shared skills to follow a Spec Kit development workflow.
- Teams documenting an existing codebase can use
speckit-baselineto derive a specification before planning changes. - Cursor Agent users can use the generated Cursor skill copies in
.cursor/skills/. - Teams using GitHub Copilot CLI can use the generated agents and prompt wrappers in
.github/agents/and.github/prompts/. - OpenCode users can invoke generated commands from
.opencode/commands/and use the workflow to review or update planned tasks.
How do you install or deploy this agent?
Clone the repository and install Spec Kit, then initialize the target project with the appropriate integration. The README does not give Spec Kit installation commands, version requirements, or runtime authentication steps. To use the shared skills in a target runtime that consumes Agent Skills directly, copy them as needed:
git clone https://github.com/dceoy/speckit-agent-skills.git
cd speckit-agent-skills
# Install Spec Kit using its documented installation method.
# Then, from the target project:
specify init --here --integration <integration>
# If the target runtime consumes Agent Skills directly:
cp -a skills/* /path/to/project/skills/How do you use this agent?
From the target project, run the initialization command and replace <integration> with the chosen integration. The repository's current CI configuration uses codex, agy, cursor-agent, copilot, opencode, and claude. Then invoke the relevant speckit-* skill through the runtime's skill flow or command; skills do not always run automatically.
specify init --here --integration <integration>What are this agent's strengths and limitations?
- The shared
skills/directory is exposed to both Claude Code and Codex CLI through directory links, allowing them to use the same skill set. - The workflow includes
speckit-baselinefor deriving specifications from existing code, as well as the main path from requirements through implementation. - CI regenerates outputs when the tracked Spec Kit version changes and is configured for several named integrations.
- Using the repository requires installing Spec Kit and initializing a target project for its chosen integration.
- Generated entry points live in different runtime-specific directories; Cursor, Copilot, and OpenCode do not all use the same layout.
- Gemini CLI commands are checked in but outside the current CI regeneration set, and legacy runtime layouts are not maintained.
How does this agent compare with similar options?
Key facts side by side with the most closely related agents.
| Agent | Source review | Form / cost | Stars | Updated | Language | Full support on |
|---|---|---|---|---|---|---|
| Spec Kit Skills for Coding Runtimes This agent | 51 · Major gaps | Agent plugin / skill | ★ 151 | 1d ago | Shell | Codex · Claude Code |
| AI-DLC Workflows | 49 · Major gaps | CLIFree + model costs | ★ 5.5k | today | TypeScript | Codex · Claude Code |
| Claudexor | 76 · Good | CLIFree + model costs | ★ 516 | today | TypeScript | Codex · Claude Code |
| EOC — Everything OpenAI Codex Workflow OS | 75 · Good | Agent plugin / skillFree + model costs | ★ 86 | 4mo ago | JavaScript | Codex |
How does FollowAgents rate this agent?
Why each dimension lost points
The workflow declares contents:read as its baseline and grants contents:write and pull-requests:write specifically to regeneration; however, Dependabot auto-merge is unconditional, and the supplied material does not explain runtime skill permissions, confirmation, data flows, sensitive-data handling, or rollback. Several Actions are pinned to commit hashes, providing some evidence for dependency security. External effects include committing generated output and automatic merges, with little recovery guidance. The README identifies Spec Kit as the upstream and explains version tracking, supporting source traceability; publisher identity and maintenance responsibility remain unclear and receive no credit.
The README's integration paths mostly align with the workflow configuration, and it notes that Gemini is outside the current regeneration set, supporting reasonable internal consistency but not full corroboration. Updates depend on the latest Spec Kit release and external runtime tools; prerequisites are listed, but availability is not addressed. The README suggests inspecting SKILL.md and project structure when a skill fails, but gives no concrete error messages or troubleshooting examples, so failure-message support is thin.
The README describes a spec-driven workflow, optional steps, and several Agent runtimes, covering common scenarios. It also says skills may not auto-run and should be invoked explicitly, and documents differing integration layouts. Trigger guidance is general, and the supplied material does not establish when each individual skill applies, so trigger precision is limited. Environment guidance is relatively clear, but the skill contents are not supplied to establish broader adaptability.
The README organizes information with a directory tree, integration table, workflow diagram, and regeneration instructions. The quickstart, prerequisites, usage notes, and legacy-layout guidance are useful, but there is no FAQ or concrete skill example. The full AGPL-3.0 license text is present, and VERSION.md plus the update workflow provide version tracking; there is no changelog, and maintenance ownership is not explicit, so those criteria do not receive full marks.
The README presents a workflow from principles and requirements through planning, tasks, and implementation, along with checklist and task-to-issue utilities, indicating potential value. It supplies no output examples, quality evidence, or cost information, so output usability and cost-benefit cannot be established and score low.
Key claims about layout, integrations, and regeneration can be traced to the README, and the workflow files support some of the integration configuration. Independent corroboration across files is limited, and no skill bodies, test results, or execution evidence are supplied. The documentation generally presents configuration and advice as such, but the evidence is insufficient to verify all product behavior, so full marks are not justified.
- The supplied material omits the skill bodies, so their instructions, specific permissions, and output behavior cannot be reviewed statically.
- CI regeneration can write to the repository and create or update pull requests; a separate workflow configuration enables unconditional Dependabot auto-merge. Review the target repository and merge conditions before relying on these workflows.
FAQ
Is this a hosted service or a set of skills to add to a project?
specify init, or copy skills when the runtime requires it.Which runtimes have documented integrations?
codex, agy, cursor-agent, copilot, opencode, and claude. Gemini CLI commands are checked in, but CI does not currently regenerate them.Do the skills run automatically?
What should I check if a skill fails?
SKILL.md and verify the required Spec Kit project structure and prerequisites.