Custom Claude Code plugins for autonomous agent orchestration. Each plugin is independently installable.
CHOICES.md management — CRUD with cascading impact review, structural audit, and bootstrap from existing docs.
Skills:
/choose-wisely:install- Bootstrap CHOICES.md for this project/choose-wisely:choose- Add/update/remove architectural decisions/choose-wisely:choose-init- Bootstrap CHOICES.md from existing docs/choose-wisely:choose-audit- Check for contradictions and structural issues
Planning gate verification — validate PLAN.md, run design review, handle replans.
Skills:
/verify:install- Bootstrap PLAN.md for this project/verify:plan- Validate PLAN.md structure (phases, steps, success gates)/verify:design- Architectural review with 9 parallel subtasks/verify:replan- Trigger visible replan on scope changes
Agents: analyzer, checker, planner, reviewer, security-analyzer
Discord integration — outbox relay, inbox processing, thread state, secret prompting, dashboard posting, and trust management.
Skills:
/discord:install- Bootstrap Discord integration (inbox/outbox, trust configs, hooks)/discord:thread-title- Set thread color and 5-word title via outbox/discord:secret-request- Prompt user for env vars via Discord ephemeral message/discord:dashboard- Aggregate and post metrics to Discord
Agents: dashboard
Lib: outbox.py, inbox.py, process-inbox.py, inbox-notify.py
Hooks: check-inbox.py (PreToolUse — processes inbox before each tool call)
PULSE.md-based scheduling, health monitoring, and container boot initialization.
Skills:
/scheduler:install- Bootstrap PULSE.md, health config, and scheduler state/scheduler:pulse- Manage scheduled tasks (list/add/remove/edit/run)
Lib: director.py (health+pulse asyncio loops), healthMon.py, boot.py, permissions.py, deny-check.py
Autonomous project orchestration loop — the core state machine that drives iterative development.
Skills:
/director:init- Bootstrap all plugins in order (thin orchestrator)/director:install- Bootstrap director core files + scan+archive cleanup/director:beat- Single state machine tick (assess → plan → implement → verify)/director:resume- Orient from saved state files at session start/director:stop- Clean stop with handoff notes/director:status- Show current state and progress/director:handoff- Session handoff with MEMORY.md update/director:agentic- Guide for agentic workflow patterns/director:director- Main orchestration loop controller
Agents: Director (main), developer (task executor), explorer (read-only research), upskill (pattern extraction)
choose-wisely — standalone (no deps)
verify — reads CHOICES.md (soft dep on choose-wisely)
discord — standalone (reads/writes outbox/inbox files)
scheduler — standalone (runs container loops)
director — orchestrates the above; all others are optional enhancements
Install only what you need. A team that just wants CHOICES.md management installs only choose-wisely.
{
"plugins": {
"local_plugins": [
"~/cc-plugins/choose-wisely",
"~/cc-plugins/verify",
"~/cc-plugins/discord",
"~/cc-plugins/scheduler",
"~/cc-plugins/director"
]
}
}Run /director:init (or --with-discord for Discord support) to bootstrap all templates.
See docs/ for Claude Code configuration best practices.