Add Feature — Skill Index and Usage
This skill is the entry point for project and feature workflows. Use it to choose the right child skill or to see the full catalog.
Meta child skills (invoke by name)
| Skill |
When to use |
| add-spec |
User says "spec a project", "create a spec", "spec this idea", "plan a project", "write a spec". Runs full brainstorming workflow then writes a no-frills spec to docs/specs/{name}.md for Ralph loops and add-project. |
| add-project |
User says "build project", "new project", "create project", "scaffold project". Creates project shell + Ralph execution loop (Phase 1 delegates to add-spec for the spec); does not scaffold Next.js (Ralph does that at step 1.1). |
| improve-skill |
User says "polish a skill", "refine skill", "test a skill", "improve skill docs", "validate a skill". Iteratively implements one skill in a sandbox, validates, reflects on errors, updates SKILL.md, reverts, retries until clean. |
| test-skills |
User says "test all skills", "validate skill composition", "integration test skills". Applies all skills in dependency order across sandboxes, validates composition, fixes markdown, ports back. |
| eval-projects |
User says "eval projects", "evaluate loops", "review ralph loops", "suggest loop improvements". Reviews evaluation tracking from all ralph loops and logs in loops/, then suggests critical and nice-to-have improvements. |
How to use child skills
- Explicit invocation: Invoke by name (e.g.
/skills:add-project or /skills:create-next). The agent loads that skill's full instructions.
- From this index: If the user asks to "add a feature", "build a project", or "spec a project", read this SKILL.md then invoke the appropriate child (add-spec for specs only, add-project for full build, improve-skill, test-skills, or the specific feature skill).
- Paths: All skills live under
skills/<name>/SKILL.md. References in add-project, improve-skill, and test-skills use skills/ as the canonical root.
Full skill catalog
The plugin provides the following skills. Invoke any by name when the task matches that skill's scope.
Meta & project
- add-feature (this skill)
- add-spec
- add-project
- improve-skill
- test-skills
- eval-projects
- create-next
- docker
- env-config
- setup-vercel
- setup-lefthook
- env-from-1password
- ralph-ci
- openclaw-generator
UI & design
- add-shadcn
- add-seo
- add-pwa
- motion
- lottie
- react-flow
- react-three-fiber
- image-editor
Data & backend
- db
- storage
- storage-ui
- queue
- auth
- auth-dev
- email
- payments
- search
- cms
- compliance
- observability
- analytics
AI
- ai-sdk
- ai-core
- ai-chat
- ai-tools
- ai-reasoning
- ai-memory
- ai-tasks
- ai-artifacts
- ai-mcp
- ai-image-gen
- ai-video-gen
- ai-generative-ui
- ai-captions
- ai-meeting-notes
- ai-voice-room
- ai-rag-ingest
- ai-rag-vectors
- ai-rag-viewer
- ai-rag-chat
- ai-rag-app
- ai-multi-agent
Realtime & media
- realtime
- video-room
- video-ui
- video-player
- screen-share
- audio-room
- stream-mux
- livestream
- recording
- transcription
- video-messaging
- media-bunny
Debrief & content
- debrief
- debrief-intake
- debrief-index
- debrief-viewer-sync
- debrief-cited-chat
- debrief-deep-analysis
- debrief-executive-summary
- debrief-export
- debrief-demo-access
Other
- group-chat
- notifications-push
- knowledge-sync
- voice-retell
- embeddable-widget
- structured-editor
- slides
- spec-export
- workflow
- rss
- yt-dlp
- dj-tracklist-downloader
- unicorn-studio
- mcp-server
Testing
- e2e
- e2e-chat
- e2e-streaming
- e2e-video
Dependencies
Child skills declare dependencies in their frontmatter (metadata.dependencies or dependencies). When applying multiple skills, respect dependency order (e.g. create-next before add-shadcn; db before auth). add-project and test-skills build dependency graphs from skill frontmatter.
1---2name: add-feature3description: Parent skill for feature and project workflows. Use when the user wants to build a project, add a feature, polish a skill, test all skills, or browse the skill catalog. Routes to add-project (build project from idea), improve-skill (refine one skill in a sandbox), test-skills (validate full skill composition), or any child skill by name.4---56# Add Feature — Skill Index and Usage7<!-- markdownlint-disable MD060 -->89This skill is the entry point for project and feature workflows. Use it to choose the right child skill or to see the full catalog.1011## Meta child skills (invoke by name)1213| Skill | When to use |14|-------|-------------|15| **add-spec** | User says "spec a project", "create a spec", "spec this idea", "plan a project", "write a spec". Runs full brainstorming workflow then writes a no-frills spec to docs/specs/{name}.md for Ralph loops and add-project. |16| **add-project** | User says "build project", "new project", "create project", "scaffold project". Creates project shell + Ralph execution loop (Phase 1 delegates to add-spec for the spec); does not scaffold Next.js (Ralph does that at step 1.1). |17| **improve-skill** | User says "polish a skill", "refine skill", "test a skill", "improve skill docs", "validate a skill". Iteratively implements one skill in a sandbox, validates, reflects on errors, updates SKILL.md, reverts, retries until clean. |18| **test-skills** | User says "test all skills", "validate skill composition", "integration test skills". Applies all skills in dependency order across sandboxes, validates composition, fixes markdown, ports back. |19| **eval-projects** | User says "eval projects", "evaluate loops", "review ralph loops", "suggest loop improvements". Reviews evaluation tracking from all ralph loops and logs in loops/, then suggests critical and nice-to-have improvements. |2021## How to use child skills2223- **Explicit invocation:** Invoke by name (e.g. `/skills:add-project` or `/skills:create-next`). The agent loads that skill's full instructions.24- **From this index:** If the user asks to "add a feature", "build a project", or "spec a project", read this SKILL.md then invoke the appropriate child (add-spec for specs only, add-project for full build, improve-skill, test-skills, or the specific feature skill).25- **Paths:** All skills live under `skills/<name>/SKILL.md`. References in add-project, improve-skill, and test-skills use `skills/` as the canonical root.2627## Full skill catalog2829The plugin provides the following skills. Invoke any by name when the task matches that skill's scope.3031### Meta & project3233- add-feature (this skill)34- add-spec35- add-project36- improve-skill37- test-skills38- eval-projects39- create-next40- docker41- env-config42- setup-vercel43- setup-lefthook44- env-from-1password45- ralph-ci46- openclaw-generator4748### UI & design4950- add-shadcn51- add-seo52- add-pwa53- motion54- lottie55- react-flow56- react-three-fiber57- image-editor5859### Data & backend6061- db62- storage63- storage-ui64- queue65- auth66- auth-dev67- email68- payments69- search70- cms71- compliance72- observability73- analytics7475### AI7677- ai-sdk78- ai-core79- ai-chat80- ai-tools81- ai-reasoning82- ai-memory83- ai-tasks84- ai-artifacts85- ai-mcp86- ai-image-gen87- ai-video-gen88- ai-generative-ui89- ai-captions90- ai-meeting-notes91- ai-voice-room92- ai-rag-ingest93- ai-rag-vectors94- ai-rag-viewer95- ai-rag-chat96- ai-rag-app97- ai-multi-agent9899### Realtime & media100101- realtime102- video-room103- video-ui104- video-player105- screen-share106- audio-room107- stream-mux108- livestream109- recording110- transcription111- video-messaging112- media-bunny113114### Debrief & content115116- debrief117- debrief-intake118- debrief-index119- debrief-viewer-sync120- debrief-cited-chat121- debrief-deep-analysis122- debrief-executive-summary123- debrief-export124- debrief-demo-access125126### Other127128- group-chat129- notifications-push130- knowledge-sync131- voice-retell132- embeddable-widget133- structured-editor134- slides135- spec-export136- workflow137- rss138- yt-dlp139- dj-tracklist-downloader140- unicorn-studio141- mcp-server142143### Testing144145- e2e146- e2e-chat147- e2e-streaming148- e2e-video149150## Dependencies151152Child skills declare dependencies in their frontmatter (`metadata.dependencies` or `dependencies`). When applying multiple skills, respect dependency order (e.g. create-next before add-shadcn; db before auth). add-project and test-skills build dependency graphs from skill frontmatter.