Current generated skill index

This page indexes the product-managed skills embedded in the current SpecBind CLI. specbind install renders the same agent-neutral body for each selected agent while preserving the platform-specific front matter and invocation syntax.

For the files installed or maintained by these skills, see the current generated artifact index. For the design history behind the set, see the target skill catalog and Decision 0176.

Every supported Agent profile receives the same 15 skills:

• Claude Code: .claude/skills/<skill>/SKILL.md; invoked as /sb-*
• Codex: .agents/skills/<skill>/SKILL.md; invoked as $sb-*
• Generic: .agents/skills/<skill>/SKILL.md; invocation is defined by the compatible host rather than by SpecBind

Selecting both Codex and generic installs each shared .agents/skills/ target once.

Codex installations also receive .agents/skills/<skill>/agents/openai.yaml. It presents branded names such as SpecBind Plan, a compact UI description, and an example prompt that names the exact $sb-* identifier. This OpenAI-specific metadata is not installed for Claude Code or the generic profile. It does not change implicit invocation or declare tool dependencies. See Decision 0183.

Skill	Current role
sb-configure	Review and change supported SpecBind project configuration, coordinate the owning workflow, verify the result, and complete authorized aftercare.
sb-discovery	Confirm milestone scope from a request, explicit local Source Collection, or selected existing implementation; classify durable boundaries, delegate state changes to the CLI, and author provenance-bearing Roadmaps, Briefs, and adoption Research handoffs.
sb-plan	The only planning entry point: take one named Spec or every Spec-backed milestone item through Tasks approval, or run one explicitly requested Requirements, Design, or Tasks phase for one named Spec.
sb-drive	Drive the active milestone through safe reachable planning, implementation, and validation work, park branch-local attention, and stop before release execution.
sb-gap-analysis	Compare intended work with the repository and preserve useful milestone-local Research without becoming a gate.
sb-validate-design	Independently judge Design coverage, boundaries, buildability, self-containment, and architectural fit.
sb-contract-review	Review the milestone's complete persistent Contract graph and accept the review required before Tasks authoring.
sb-implement	Implement one Spec-backed or Direct Roadmap item using the required dispatched implementation and review cycle.
sb-review-task	Judge one implemented Task from its actual diff and approved inputs without applying fixes.
sb-debug	Establish and categorize the root cause of a stopped run and return a bounded next action without applying it.
sb-validate-implementation	Judge one Spec's implementation against its active Requirement IDs and accept completion evidence only on GO.
sb-verify-completion	Check an explicit completion claim against fresh evidence without changing lifecycle state.
sb-release	Bind the release, execute project release guidance, verify the result, and finalize the complete milestone.
sb-status	Explain current Spec, milestone, or task state and the next available action without judging completion.
sb-steering	Bootstrap, synchronize, repair, or add durable project guidance.

There are no compatibility aliases for earlier kiro-*, any former specbind-* product Skill, removed phase-specific specbind-plan-*, or specbind-adopt-existing Skill names. Milestone and Spec initialization and existing-implementation adoption are routed through sb-discovery, not separate skills.

Sources of truth

• Agent-neutral skill sources: tools/specbind/assets/skills/
• Registry and per-agent rendering: tools/specbind/src/catalog/skill.rs
• Installation planning and refresh: tools/specbind/src/installation/install.rs
• Mechanical conformance tests: tools/specbind/tests/skill.rs and tools/specbind/tests/cli.rs
• Behavioral verification index: docs/skill-forward-tests.md
• Measurement dashboard: docs/skill-forward-tests/results.md
• Historical run records: docs/skill-forward-tests/runs/
• Findings worklist: docs/skill-forward-tests/findings.md

When a product skill changes, update its one embedded package and the applicable mechanical and forward tests. Each package has one SKILL.md entrypoint and may have directly linked references/ files for conditional detail. Both agent renderings are derived from that package.