Stats
Actions
Tags
From selfwork
Start or resume selfwork orchestration. Reads active run to resume execution; if no active run, clarifies requirements and initializes a new run.
How this command is triggered — by the user, by Claude, or both
Slash command
/selfwork:selfworkThe summary Claude sees in its command listing — used to decide when to auto-load this command
Use the `selfwork` skill to execute the start/resume flow.
## Startup
Before any other action:
1. Run bootstrap to initialize directories and get the run context:
2. Confirm the result shows the current project's `.claude/selfwork/` as the state root.
3. If an active run exists (`active` file is present), read `state.json` and resume from current `status`.
4. If no active run exists, bootstrap creates one — proceed with the Clarify or Plan phase.
## Root Separation
- `${CLAUDE_PLUGIN_ROOT}` — plugin assets only (commands, agents, skills, scripts). Read-only.
- `./.claude/selfwork/`...Use the selfwork skill to execute the start/resume flow.
Before any other action:
bun "${CLAUDE_PLUGIN_ROOT}/skills/selfwork/scripts/bootstrap.ts"
.claude/selfwork/ as the state root.active file is present), read state.json and resume from current status.${CLAUDE_PLUGIN_ROOT} — plugin assets only (commands, agents, skills, scripts). Read-only../.claude/selfwork/ — runtime state for the current project. All state writes go here.${CLAUDE_PLUGIN_ROOT}.Follow the selfwork skill's phase model:
The CEO never writes code, runs tests, or performs reviews directly. All implementation work is delegated to subagents.
npx claudepluginhub royisme/agent-skills --plugin selfwork/helpDisplays Orchestrator documentation including commands list, quick start guide, development lifecycle diagram, specialized agents, quality gates, error recovery, and state files.
/configInteractively configures task prompt, execution mode, models, and approval settings for /orchestrate:start, then launches or displays the command.
/executeExecutes tasks from a completed plan using isolated subagents. Prompts for project directory, verifies planning, runs task-executor subagents, handles checkpointing and crash recovery.
/supervisor-driveOrchestrates autonomous project workflows: surveys state, selects and runs the appropriate engine, re-evaluates, and loops. Supports interactive, auto, and plan-only modes with safety gates and budget caps.