| name | workspace-orchestration |
| description | Use when working in a persistent-slot workspace that uses docko. Covers slot claims,
delegated teammate inheritance, and the rule that code edits inside slots/ require an active claim.
|
docko Workspace Orchestration
Use this skill whenever work touches files in slots/ or when the user asks about workspace coordination.
Quick Path
- Start at the workspace root.
- Run
/dock-status or docko status --root . --brief.
- Prefer
docko slot acquire --root . --branch <branch> --task "<task>" --brief when you want docko to choose the next available slot using round-robin selection.
- If every slot is busy and docko asks whether it should create a fresh managed clone, answer explicitly.
- Use
docko claim --root . --resource slot --id <slot> --branch <branch> --task "<task>" only when you already know the exact slot.
- Do the work inside that slot.
- Release the slot when done:
docko release --root . --resource slot --id <slot>
Prefer slash commands when installed:
/dock-status
/dock-claim <slot> <branch> <task>
/dock-heartbeat <slot>
/dock-release <slot>
Rules
- Work from the root. Edit code in
slots/*.
- If a parent session already owns the slot, reuse that authority instead of creating a second claim.
- If docko reports
AMBIGUOUS_SESSION, retry with an explicit --session <id> from docko session list --root . --brief; do not end existing sessions unless the user asked for cleanup.
- If every slot is busy and the user already approved the fallback, add
--clone-when-busy to docko slot acquire.
- If
docko is not runnable, check DOCKO_BIN. If it still fails, stop and tell the user.
- Do not inspect slots one by one or replace the CLI with
docko/registry.json during normal work.