notque/vexjoy-agent / skills/process/workflow
workflow
VerifiedStructured work: multi-phase tasks, feature builds, planning, objective loops, hill climbing.
references/project-coordinator/agent-capability-map.md
View on GitHub# Worker Capability Map Reference
> **Scope**: Task-to-reference routing for coordinated workers; capabilities and boundaries per domain.
Every worker is a fresh `general-purpose` agent (Agent tool `subagent_type: "general-purpose"`). The domain lives in a skill reference, not in an agent. Start each worker prompt with two instructions: call the Skill tool with the row's skill, then `Read ~/.claude/skills/<skill>/references/<file> and follow it; load every file its Reference Loading Table matches.`
---
## Primary Routing Table
| Task Type | Skill | Reference to load | Avoid |
|-----------|-------|-------------------|-------|
| Go source code | `programming` | `references/golang-general.md` | a worker with no reference |
| TypeScript backend API | `programming` | `references/nodejs-api.md` (runtime bugs: `references/typescript-debugging.md`) | frontend reference |
| TypeScript frontend/React | `frontend` | `references/typescript-frontend.md` | backend reference |
| Python scripts/data | `programming` | `references/python-general.md` (OpenStack: `references/python-openstack.md`) | a worker with no reference |
| Database schema/migrations | `data` | `references/database.md` | application references |
| Kubernetes/Helm | `kubernetes` | `references/kubernetes-helm.md` | `references/ansible-automation.md` |
| Ansible playbooks | `kubernetes` | `references/ansible-automation.md` | `references/kubernetes-helm.md` |
| OpenSearch/Elasticsearch | `kubernetes` | `references/opensearch-elasticsearch.md` | `data` references |
| Prometheus/Grafana | `kubernetes` | `references/prometheus-grafana.md` | — |
| Swift iOS/macOS | `programming` | `references/swift-general.md` | — |
| Kotlin Android/JVM | `programming` | `references/kotlin-general.md` | — |
| PHP backend | `programming` | `references/php-general.md` | — |
| React Native / Expo | `frontend` | `references/react-native.md` | `references/typescript-frontend.md` |
| Code review | `review` | `references/reviewer-code.md` (fallback `references/reviewer-system.md`) | — |
| Security audit | `review` | `references/reviewer-system.md` | — |
| Performance | `frontend` | `references/performance-optimization.md` | — |
| Documentation | `writing` | `references/technical-documentation.md` | — |
`security` is a skill: call it with the Skill tool for a standalone scan of git changes.
Verify the reference exists before dispatch: `ls ~/.claude/skills/{skill}/references/{file}`
---
## Domain Scope Boundaries
**Go worker** (`programming`, `golang-general.md`): CAN modify `.go` files, run `go build/test/vet`, add packages, fix linting. CANNOT modify non-Go files, make architectural decisions, run migrations.
**Node API worker** (`programming`, `nodejs-api.md`): CAN do REST endpoints, middleware, auth, webhooks, DB integration. CANNOT do frontend components, native code, infrastructure.
**Database worker** (`data`, `database.md`): CAN do schema design, migrations, indexes, query optimization. CANNOT do application code. **Mandatory sequencing**: the database worker ALWAYS runs before application workers on schema changes.
**Performance worker** (`frontend`, `performance-optimization.md`): CAN do Core Web Vitals, bundle analysis, rendering optimization, profiling. CANNOT fix unrelated logic bugs.
---
## Compound Task Routing
### API + Frontend Feature
```
1. database worker (data/database.md) → schema migration
2. Node API worker → endpoint (parallel with 3 if no shared types)
3. frontend worker (frontend/typescript-frontend.md) → UI
4. code review worker (review/reviewer-code.md) → full-stack review
```
### Go Service Performance
```
1. performance worker (frontend/performance-optimization.md) → profile, identify hotspot (read-only)
2. Go worker (programming/golang-general.md) → fix
3. Go worker → benchmark validation
```
### Infrastructure + Application
```
1. Helm worker (kubernetes/kubernetes-helm.md) → Helm chart (parallel with 2 if isolated)
2. Ansible worker (kubernetes/ansible-automation.md) → provisioning
3. Application worker → smoke test
```
---
## Preferred Patterns
### Name the Reference in Every Dispatch
**Signal**: A `general-purpose` worker dispatched for Go errors with no skill or reference in its prompt.
**Fix**: Look up the task type in the routing table; put the Skill call and reference read line first in the prompt.
### Match Reference to Language Domain
**Signal**: TypeScript backend work sent with the frontend reference.
**Fix**: Check imports — `express`/`@nestjs`/`fastify` → Node API worker. `react`/`next` → frontend worker.
```bash
rg "import.*express|import.*@nestjs" --files-with-matches src/ # → backend
rg "import.*react|import.*next" --files-with-matches src/ # → frontend
```
### Route Schema Changes to the Database Worker
**Signal**: Asking the Node API worker to add a column.
**Fix**: Database worker first for the migration, then the application worker for ORM models.
---
## Worker Selection Decision Tree
```
.go files? → programming + golang-general.md
TypeScript backend/API? → programming + nodejs-api.md
TypeScript frontend/React? → frontend + typescript-frontend.md
Database schema? → data + database.md (first, before app workers)
K8s/containers? → kubernetes + kubernetes-helm.md
VM provisioning? → kubernetes + ansible-automation.md
Performance? → frontend + performance-optimization.md
Code review? → review + reviewer-code.md
Otherwise? → find the skill whose references match → general-purpose with that skill
```
Related skills
| Skill | Stars | Updated | Works with | Status | Install |
|---|---|---|---|---|---|
| planning-and-task-breakdownaddyosmani/agent-skillsBreaks work into ordered tasks. Use when you have a spec or clear requirements and need to break work into implementable tasks. Use when a task feels too large to start, when you need to estimate scope, or when parallel work is possible. | Stars: 119k | Updated | Claude CodeCodex | Verified | Install |
| screenpipethedotmack/claude-memSearch screen and audio history recorded by Screenpipe and optionally save selected context into claude-mem. Use when the user asks to recall desktop activity, a meeting, or something seen in another app through Screenpipe. | Stars: 118k | Updated | Claude CodeCodex | Verified | Install |
| contact-widgetnexu-io/open-designSelf-contained floating chat widget with welcome screen, social links, meeting button, and message input. Single HTML file, zero dependencies. | Stars: 108k | Updated | Claude CodeCodex | Verified | Install |
| meeting-notesnexu-io/open-designMeeting notes page — title bar with attendees, agenda checklist, decisions block, action items table with owners + dates, and a "next meeting" footer. Use when the brief mentions "meeting notes", "minutes", "1:1 notes", "all-hands recap", or "会议纪要". | Stars: 108k | Updated | Claude CodeCodex | Verified | Install |