When to invoke
Use when the task clearly wants a vibestack workflow but the right one is not
obvious, or when the host has no slash-command picker and the pack has to be
found by name. In Claude Code, Cursor and Kiro, skills are /name. In Codex
there is no / picker — reference a skill as $name inside an ordinary
message (Use $office-hours to shape this idea).
/vibe — pick the right skill
The router is named vibe, not vibestack: the pack's own checkout is
sometimes parked at ~/.claude/skills/vibestack, and a skill installing to that
path would collide with it — the installer's own integration test asserts that
checkout survives.
Route on intent. Process skills come first: brainstorm or plan before
implementing, investigate before fixing, review before shipping.
| When the task is… |
Use |
| Shape a rough idea into a design doc |
office-hours |
| Turn intent into a precise, executable spec |
spec |
| Plan a feature, refactor, or architecture change |
plan-eng-review, autoplan |
| Weigh product scope or the bigger problem |
plan-ceo-review |
| Plan a UI change, or review a design before it is built |
plan-design-review |
| Pressure-test the developer experience of a plan |
plan-devex-review |
| Debug an error, test failure, or odd behavior |
investigate |
| Review a diff before merge |
review |
| Get a second opinion from a different model |
codex, claude |
| Audit security |
cso |
| QA a running web app |
qa (fixes), qa-only (report) |
| Drive a browser, scrape a page, pair a remote agent |
browse, scrape, open-browser, pair-agent |
| Review a shipped UI, or explore design directions |
design-review, design-shotgun, design-consultation |
| Ship: tests, version, changelog, PR |
ship |
| Address PR review threads and failing CI |
address-pr-review |
| Merge, deploy, and confirm production health |
land-and-deploy, canary |
| Update docs after shipping |
document-release, document-generate |
| Strip machine-sounding prose |
unslop |
| Check code-quality health or find refactors |
health, improve-arch |
| Review AWS spend |
aws-cost |
| Cap paid-inference spend |
ai-cost-guard |
| Audit Bedrock guardrails, region and tenant isolation |
bedrock-guardrails |
| Review a knowledge base or RAG pipeline |
kb-review |
| Review an Amazon Connect, Lex or voice solution |
connect-review |
| Evaluate an agent or prompt |
agent-eval |
| Review an MCP server |
mcp-review |
| Save or restore working context across sessions |
context-save, context-restore |
| Guard a risky session |
careful, freeze, guard (and unfreeze to release) |
| Update the pack itself |
vibe-upgrade |
Anything not listed here is in the full index — read it rather than guessing at
a skill name:
cat "${CLAUDE_SKILL_DIR:-$HOME/.claude/skills/vibe}/skills-index.md" 2>/dev/null \
|| ls ~/.claude/skills/
Two rules when routing:
- Pick one. Don't stack skills; each carries its own full workflow, and a
skill's own instructions take precedence once invoked.
- Don't guess a name. If nothing above fits, say so and proceed normally —
these are workflows, not dependencies.
1---2name: vibe3description: Router for the vibestack skill suite — name the task and it points you at the right skill. Use when you know vibestack is installed but not which of its skills fits, or when an agent has no slash-command picker (Codex) and needs to find the pack by name.4---56## When to invoke78Use when the task clearly wants a vibestack workflow but the right one is not9obvious, or when the host has no slash-command picker and the pack has to be10found by name. In Claude Code, Cursor and Kiro, skills are `/name`. In Codex11there is no `/` picker — reference a skill as `$name` inside an ordinary12message (`Use $office-hours to shape this idea`).1314# /vibe — pick the right skill1516The router is named `vibe`, not `vibestack`: the pack's own checkout is17sometimes parked at `~/.claude/skills/vibestack`, and a skill installing to that18path would collide with it — the installer's own integration test asserts that19checkout survives.2021Route on intent. Process skills come first: brainstorm or plan before22implementing, investigate before fixing, review before shipping.2324| When the task is… | Use |25|---|---|26| Shape a rough idea into a design doc | `office-hours` |27| Turn intent into a precise, executable spec | `spec` |28| Plan a feature, refactor, or architecture change | `plan-eng-review`, `autoplan` |29| Weigh product scope or the bigger problem | `plan-ceo-review` |30| Plan a UI change, or review a design before it is built | `plan-design-review` |31| Pressure-test the developer experience of a plan | `plan-devex-review` |32| Debug an error, test failure, or odd behavior | `investigate` |33| Review a diff before merge | `review` |34| Get a second opinion from a different model | `codex`, `claude` |35| Audit security | `cso` |36| QA a running web app | `qa` (fixes), `qa-only` (report) |37| Drive a browser, scrape a page, pair a remote agent | `browse`, `scrape`, `open-browser`, `pair-agent` |38| Review a shipped UI, or explore design directions | `design-review`, `design-shotgun`, `design-consultation` |39| Ship: tests, version, changelog, PR | `ship` |40| Address PR review threads and failing CI | `address-pr-review` |41| Merge, deploy, and confirm production health | `land-and-deploy`, `canary` |42| Update docs after shipping | `document-release`, `document-generate` |43| Strip machine-sounding prose | `unslop` |44| Check code-quality health or find refactors | `health`, `improve-arch` |45| Review AWS spend | `aws-cost` |46| Cap paid-inference spend | `ai-cost-guard` |47| Audit Bedrock guardrails, region and tenant isolation | `bedrock-guardrails` |48| Review a knowledge base or RAG pipeline | `kb-review` |49| Review an Amazon Connect, Lex or voice solution | `connect-review` |50| Evaluate an agent or prompt | `agent-eval` |51| Review an MCP server | `mcp-review` |52| Save or restore working context across sessions | `context-save`, `context-restore` |53| Guard a risky session | `careful`, `freeze`, `guard` (and `unfreeze` to release) |54| Update the pack itself | `vibe-upgrade` |5556Anything not listed here is in the full index — read it rather than guessing at57a skill name:5859```bash60cat "${CLAUDE_SKILL_DIR:-$HOME/.claude/skills/vibe}/skills-index.md" 2>/dev/null \61 || ls ~/.claude/skills/62```6364Two rules when routing:6566- **Pick one.** Don't stack skills; each carries its own full workflow, and a67 skill's own instructions take precedence once invoked.68- **Don't guess a name.** If nothing above fits, say so and proceed normally —69 these are workflows, not dependencies.
timurgaleev