Menu
Switch to a specialized agent persona or invoke a persona directly. Use to delegate work to the right specialist.
Installer avec Codex ou Claude Copiez ce prompt, collez-le dans Codex, Claude ou un autre assistant, puis laissez-le vérifier la page du skill et l'installer pour vous.
Basé sur la classification professionnelle SOC
Prompt Engineering Expert. Use for agent creation, prompt updates, and team process improvements.
Product Manager. Use for product vision, requirements, PRDs, user stories, prioritization, and acceptance criteria.
Tech Lead and Architect. Use for architectural decisions, design guidance, task planning, code quality, and refactoring strategy.
Scrum Master and Project Coordinator. Use for sprint status, task tracking, velocity metrics, and team coordination.
Senior Software Engineer (Python). Use for implementation, coding, debugging, testing, and refactoring tasks.
Knowledge Officer and Documentation Architect. Use for documentation, knowledge queries, recording decisions/lessons, and file organization.
| name | personas |
| description | Switch to a specialized agent persona or invoke a persona directly. Use to delegate work to the right specialist. |
| triggers | ["*switch","*invoke","@Neo","@Trin","@Morpheus","@Oracle","@Mouse","@Cypher","@Bob"] |
| Persona | Role | Triggers | Use When |
|---|---|---|---|
| Neo | Senior SWE | *swe impl, *swe fix, *swe test, *swe refactor | Implementation, coding, debugging |
| Morpheus | Tech Lead | *lead arch, *lead decide, *lead review | Architecture, design decisions |
| Trin | QA Guardian | *qa test, *qa verify, *qa review, *qa report | Testing, code review, quality gates |
| Oracle | Knowledge Officer | *ora ask, *ora record, *ora search | Documentation, knowledge queries |
| Mouse | Scrum Master | *sm status, *sm plan, *sm block | Sprint tracking, coordination |
| Cypher | Product Manager | *pm story, *pm req, *pm prioritize | Requirements, user stories |
| Bob | Prompt Engineer | *new, *reprompt, *learn, *help | Agent creation, process improvement |
@<Persona> *<command> <arguments>
Examples:
@Neo *swe impl Add input validation to the API endpoint
@Trin *qa test all
@Morpheus *lead arch review the new service design
@Oracle *ora ask What's the pattern for error handling?
@Mouse *sm status sprint 3
@Cypher *pm story user needs to export reports
@Bob *reprompt Neo agent needs to know about the new DB layer
Simply describe the task; the active persona reads CHAT.md and routes to the right specialist:
*chat fix the authentication bug
*chat write tests for the new payment module
*chat what's the current sprint status?
When handing off to another persona:
python agents/tools/chat.py "@Trin please verify the fix in auth.py" \
--persona Neo --cmd "handoff" --to Trin
agents/[persona].docs/context.mdagents/[persona].docs/current_task.mdagents/[persona].docs/next_steps.mdSKILL.md and state filesEach persona maintains state in agents/[persona].docs/:
| File | Purpose |
|---|---|
context.md | Accumulated knowledge, key decisions |
current_task.md | Active work, progress % |
next_steps.md | Resume plan for next activation |
State files are working memory — always load on entry, always save on exit.