Dooo — Parallel Execution Orchestrator

skills/j-dooo/ is the canonical, hand-edited directory for this skill, per CLAUDE.md's "The Canonical Naming Contract" (the E50 reopening of 2026-09-09, which promoted skills/j-dooo/ from generated twin to sole canonical form). The j- prefix is there for collision safety — a real directory under a distinct name, so a host tool shipping its own same-named built-in command cannot shadow it (Claude Code's native skill resolution is a literal-string, directory-name-based match; see project/documentation/skill-authoring.md's "Invocation Convention").

⚠️ scripts/generate-j-alias.sh was retired by E50_S14 and no longer exists — there is nothing to run. This file was previously generated from a bare skills/dooo/SKILL.md source; E50_S15 deleted that directory. This file is now the sole canonical, hand-edited source for this skill — edit it directly.

Instructions

1. Invoke /do

Call the /do skill to let the user select and start an implementation. /do will launch a background sub-agent to handle the implementation. Once the sub-agent is launched, /do returns control here.

After /do hands back control, mark the story/task that was just started as In Progress in its board file (update the status: field in the YAML front-matter).

Exception — a deferred unit is not started. If /do reports the unit was deferred ("deferred: pre-flight checklist needs a user decision", per skills/j-do/SKILL.md step 4.1.2), do not mark it In Progress: nothing was started and /do has already left (or reverted) it at Pending. Say so explicitly to the user — that the unit was deferred, which checklist items need their decision, and that it will be offered again once resolved — then carry on to step 2 as usual.

Task-completion cleanup is inherited, not re-implemented here. /dooo never dispatches developer sub-agents itself; every task it starts goes through this step's /do call, so each one reaches /do's own step 7 and its Task-Completion Cleanup Procedure (skills/j-do/SKILL.md ### 7.1) unmodified. No separate wiring is needed in /dooo (E37_S05_T04). If /dooo ever gains a dispatch path that bypasses /do, that path must call the procedure by name.

2. Return to the board — identify parallelisable tasks

Break check first (E44_S02). This is the boundary where /dooo decides to widen the fan-out, so it is where a requested stop takes effect: after the previous /do has handed control back, and before the board is scanned for the next set of tasks.

bash "$([ -f skills/j-break/scripts/check-break.sh ] && echo skills/j-break/scripts/check-break.sh || echo node_modules/@jenga-ai/agent/skills/j-break/scripts/check-break.sh)"

• Exit 0 — no break. Continue with the scan below, unchanged.
• Exit 10 — a break was requested. The record is on stdout as mode=<soft|hard> with_status=<true|false> requested_at=<ts> requested_by=<who>. A soft break was consumed by that call; a hard break remains. Dispatch nothing, skip the scan, and skip step 3's choice prompt entirely — offering a numbered list and then refusing to act on the pick would be worse than not asking. Print the halt report below and end the loop (step 5).
• Any other exit — the checker itself failed, which is not a stop request. Print its stderr as a warning and continue with the scan.

Never read the flag file here; branch on the exit code and use the record the script prints.

Halt report.

1. Stopped: state that a break was requested and that no new sub-agent was dispatched.
2. Still running: a break stops dispatch, not execution. Sub-agents already in flight are not cancelled and will keep running to completion, writing their own statuses afterwards. List them, so their later status writes are not a surprise: the items this /dooo session marked In Progress in step 1 whose board status has not yet reached a terminal value. (ListAgents gives authoritative liveness, but only from the host session; it is not available to a delegated agent, so do not depend on it.)
3. Mode: soft or hard.
4. Resume: /dooo again for a soft break; for a hard break, /break clear first, then /dooo.
5. Scoped rapport: only when the record says with_status=true. Run bash "$([ -f skills/j-break/scripts/render-execution-status.sh ] && echo skills/j-break/scripts/render-execution-status.sh || echo node_modules/@jenga-ai/agent/skills/j-break/scripts/render-execution-status.sh)" and print its output verbatim. If it fails, say the rapport is unavailable and why. Never improvise one from the board.

The very first /do of a session is covered by /do's own entry check, so a standing hard break also stops a /dooo that has not dispatched anything yet.

Otherwise (no break):

Run scripts/render-ranked-list.sh from the project root. It scans project/board/stories/ and project/board/tasks/ against project/todo.md and prints the eligible items as a ranked, 1-indexed <id> — <title> list (stories first, then tasks) — this is the same eligibility scan and rendering /dooo has always used (status Pending, no unresolved dependencies, directly listed in project/todo.md or — for tasks — parent story listed), now implemented once in a shared script rather than described here as inline prose (E63_S01_T01; see scripts/render-ranked-list.sh's own header for the full rule and its env-var overrides). A sibling command, /todo --ranked-list, calls the same script for a non-interactive view of this same list.

3. Present choices to the user

Relay the script's output as the numbered list (use ask_user with choices). Always append "Done" as the last option.

Example:

Which task would you like to start next?
1. Create /spinoff skill (E02_S02)
2. Training Skill Assets & Templates (E01_S06)
3. Done

If there are no eligible tasks (all remaining todos are blocked or already running), skip straight to step 4.

4. Branch on user selection

• If the user selects "Done" — exit the loop and inform the user that no more implementations will be started in this session.
• If the user selects a task — go back to step 1 (invoke /do for the selected task, then loop).

5. Loop termination

The loop ends when:

• The user selects "Done", OR
• There are no more eligible parallel tasks to offer, OR
• A break was requested (/break) — detected by the check at the top of step 2. This is a requested stop, not a completion: work may remain, and in-flight sub-agents keep running (see the halt report in step 2).