execute-fleet (Atlas Fleet parallel execution)

Orchestrator skill that runs task waves across isolated git worktrees, verifies each worker branch by evidence, merges sequentially, and ends with a single PR.

AutomationAdvanced★ 607⑂ 59AI score 7/10Last updated: Aug 14, 2026

What it does

Computes runnable task waves via compute_fleet_waves and spawns exactly one worker per chunk, with an explicit model from the returned routing map (never default).
Workers build only inside isolated worktrees/branches; the full task JSON is inlined into the worker prompt so nobody reads shared tasks.json.
Trusts the branch, not the narration: requires .atlas-ai/cdd/task-<id>.json, zero non-zero Exit status lines in evidence, and no edits to protected state files.
Merges verified branches into fleet-integration one at a time, running the project checker/build gate between merges, then recomputes waves.
Defines failure paths: one re-queue for dead workers, inbox answers for NEEDS_CONTEXT, BLOCKED bookkeeping, no forced conflict resolution.
Terminates with ship-check; on exit 0 it emits SHIP_CHECK_OK once and opens a single final PR without merging it.

Who it's for

prd-taskmaster users on the licensed Atlas Fleet premium tier who want multi-agent parallel task execution.
Team leads who need proof-based gating against agents that claim completion prematurely.
Engineers studying orchestration patterns: worktree isolation, sole-writer state, wave scheduling.

Examples

A 10-task PRD at concurrency 3: wave 1 (tasks 1–3) must merge green before wave 2 dispatches.
Worker reports DONE but no CDD card exists → completion refused, task re-queued once, then marked BLOCKED while independent tasks continue.
All waves finished → python3 skel/ship-check.py returns 0 → print SHIP_CHECK_OK once and open the PR from fleet-integration.

· · · Install guide · · ·

Try it now, no install

Paste this into Claude to use the skill without installing anything.

Read the instructions in this file and follow them to help me:
https://raw.githubusercontent.com/anombyte93/prd-taskmaster/HEAD/skills/execute-fleet/SKILL.md

What I want: (describe your task here)

If Claude can't open the link, open it yourself and paste the contents instead.

↓ If it works for you, download the ZIP below and install it. Then it runs on its own — no pasting each time.

Install in the Claude app (no terminal)

Download the ZIP with the button below.
In Claude, open Settings → Capabilities and turn on 'Code execution and file creation'. (one time)
Go to Customize → Skills → + → 'Upload a skill' and upload the ZIP.

↓ Download ZIP

Install in Claude Code

Let Claude do it — paste this into Claude Code

Install the skill I found on Claude Skill Mart.
Copy the skills/execute-fleet folder from the GitHub repo anombyte93/prd-taskmaster into my ~/.claude/skills/execute-fleet/.
When it's done, tell me in one line what this skill can do.

Install with a command instead

git clone https://github.com/anombyte93/prd-taskmaster.git /tmp/prd-taskmaster && mkdir -p ~/.claude/skills && cp -r /tmp/prd-taskmaster/skills/execute-fleet ~/.claude/skills/

⚠ This is a third-party skill. Check the source repository before installing.

Check prerequisites: the prd-taskmaster plugin, Atlas Fleet premium tier, a registered/alive atlas-launcher MCP, plus existing .taskmaster/tasks/tasks.json and .taskmaster/reports/task-complexity-report.json.
Clone the repository: git clone https://github.com/anombyte93/prd-taskmaster.git
Copy the skill into place: mkdir -p ~/.claude/skills && cp -r prd-taskmaster/skills/execute-fleet ~/.claude/skills/
Restart Claude Code and confirm execute-fleet appears in your skill list.
Do not invoke it directly — it is user-invocable: false and is triggered by the pipeline when HANDOFF selects Atlas Fleet.
Before a run, make sure git status --short is empty (fleet starts only from a committed base) and that fleet-integration is your integration branch.

View source on GitHub ↗License: MIT