Skip to content

Skills

awa provides skills that guide AI agents through specific workflow tasks. Agents may pick up the right skill from context, but for reliable results we recommend invoking them explicitly with a slash command.

Use the /awa.<skill> slash command to invoke a skill by name. This works in GitHub Copilot, Claude, Cursor, Gemini, and any other agent that supports awa prompt files:

/awa.code implement the tasks in TASK-DIFF-diff-001.md
/awa.design create a design for authentication
/awa.brainstorm how to handle rate limiting
/awa.requirements write requirements for bulk export

These follow the awa pipeline: Architecture → Features → Requirements → Design → Tasks → Code → Documentation.

Create or update ARCHITECTURE.md — the system overview, components, and constraints.

/awa.architecture Update for the changes we just discussed.
/awa.architecture Add a caching layer.

Create or update feature context documents (FEAT-*.md) — problem statement, motivation, scenarios. Non-normative; sets the stage for requirements.

/awa.feature Describe bulk export — motivation and scenarios.
/awa.feature Create a FEAT doc for the new authentication flow.

Create or update requirements (REQ-*.md) in EARS format (INCOSE). Each requirement gets an ID and testable acceptance criteria.

/awa.requirements Write requirements for bulk export with retry support.
/awa.requirements Create REQ and acceptance criteria for rate limiting.

Create or update design documents (DESIGN-*.md) — components, interfaces, correctness properties. Links back to requirements via IMPLEMENTS/VALIDATES.

/awa.design Design the implementation for authentication.
/awa.design Create a design doc covering DIFF-7 through DIFF-10.

Break requirements and designs into step-by-step implementation tasks (TASK-*.md). Each task references specific acceptance criteria.

/awa.tasks Create tasks for the authentication design.
/awa.tasks Break DESIGN-DIFF into implementation work items.

Implement code and tests with full traceability markers (@awa-component, @awa-impl, @awa-test). Works from tasks, designs, or requirements.

/awa.code Implement the tasks in task 006.
/awa.code Write code for the rate limiter design.
/awa.code Implement DIFF-7 acceptance criteria.

Create or update user-facing documentation (README.md, docs/). Derives content from specs — summarizes rather than duplicates.

/awa.documentation Update README and docs to match current CLI options.
/awa.documentation Write docs for the new template testing feature.

Create concrete usage examples (EXAMPLE-*.md) — code samples, CLI demonstrations, configuration snippets for a feature.

/awa.examples Create usage examples for the overlay feature.
/awa.examples Show CLI examples for multi-target configuration.

Explore ideas, generate options, and evaluate trade-offs. Diverges first (many ideas), then converges (recommended approach).

/awa.brainstorm How to add a PKI flow for the bobcat server.
/awa.brainstorm Options for handling concurrent template generation.
/awa.brainstorm Trade-offs between polling and webhooks?

Create ad-hoc plans (PLAN-*.md) for work that doesn’t need full requirements and design. Good for quick iterations.

/awa.plan Create a plan to add a --delete flag to the CLI.
/awa.plan Plan out the migration from Biome to ESLint.

Run awa check, analyze the output, fix all errors, and re-run until clean. Handles orphaned markers, broken cross-refs, schema violations, and uncovered acceptance criteria.

/awa.check Check traceability and fix any errors.
/awa.check Run awa check and resolve all findings.

Validate that one artifact aligns with another — e.g. design covers all requirements, or code matches the design. Produces an alignment report (ALIGN-*.md).

/awa.align Align DESIGN-DIFF with REQ-DIFF.
/awa.align Check whether code fully covers DIFF-7 acceptance criteria.

Refactor code while preserving all traceability markers and behavior. Runs awa check afterward to verify nothing broke.

/awa.refactor Refactor src/core/template-resolver.ts for readability.
/awa.refactor Extract the validation logic into a separate module.

Take an idea from concept to working implementation in one pass, flowing through the full workflow. Pauses at each stage transition for confirmation (unless you request autonomous mode).

/awa.vibe Template caching with TTL support.
/awa.vibe Implement rate limiting from scratch.

Upgrade existing spec files to match current schema definitions. Useful after schema changes to bring old specs into compliance.

/awa.upgrade Upgrade the specs to match current schemas.
/awa.upgrade Migrate all REQ files to the new schema format.

Answer questions about awa itself — how to use it, configure it, or understand its concepts.

/awa.usage How do I set up awa check in CI?
/awa.usage Explain how traceability markers work.
/awa.usage What does the --overlay flag do?

Reorganize specifications logically — reorder sections, fix numbering, clean up structure without changing meaning.

/awa.spec-tidy Tidy the specs — they've got messy and out of hand!
/awa.spec-tidy Reorganize the requirements by topic.

Merge two feature codes into one — combines spec files, recodes traceability IDs, and cleans up. Uses awa spec merge under the hood.

/awa.spec-merge Merge the CHK feature code into CLI.
/awa.spec-merge Combine SRC and TGT feature codes.

Retire requirement IDs by moving them to the tombstone file (DEPRECATED.md). Removes traceability markers from code and cleans up spec references.

/awa.spec-deprecate Deprecate DIFF-3 and DIFF-4 — they're no longer needed.
/awa.spec-deprecate Retire the old authentication requirements.