A tool for running approved workflow phases in parallel, with each phase isolated in its own Git worktree.

Use it to launch approved phase workers, inspect fleet status, stop workers, preserve their branches and worktrees, and explicitly merge completed work.

It lets independent pieces of work proceed separately while keeping their files and branches isolated. It also provides status checks and controlled teardown of phase workers.

flow-fleet — Skill for Claude Code and Codex by pwdev-solucoes

Install

Getting it into your agent

One page per mod, every tool's command on it. A separate URL per tool would split the same page into five that compete with each other.

Any agent

npx skills add pwdev-solucoes/pwdev-claude-marketplace --skill flow-fleet

Clone the repo

git clone --depth 1 https://github.com/pwdev-solucoes/pwdev-claude-marketplace

Made for: Claude Code, Codex.

Or install pwdev-flow, the plugin that ships this one along with the rest of its 17 skills, 1 command.

Wrote this? Show the measurements

A badge with what this costs and how it scanned, read live from this page, so it follows the numbers instead of freezing them. Markdown for a README, HTML for a documentation site or a project page.

README.md

[![agentmods](https://agentmods.dev/badge/skills/pwdev-solucoes/pwdev-claude-marketplace/flow-fleet/github.svg)](https://agentmods.dev/skills/pwdev-solucoes/pwdev-claude-marketplace/flow-fleet)

Your own site

<a href="https://agentmods.dev/skills/pwdev-solucoes/pwdev-claude-marketplace/flow-fleet"><img src="https://agentmods.dev/badge/skills/pwdev-solucoes/pwdev-claude-marketplace/flow-fleet/github.svg" alt="Measured on agentmods" height="20"></a>

Or the 80×15 button, for a site that already has a row of RSS and ATOM ones. Only the verdict fits; the numbers stay here.

Your own site · 80×15

<a href="https://agentmods.dev/skills/pwdev-solucoes/pwdev-claude-marketplace/flow-fleet"><img src="https://agentmods.dev/badge/skills/pwdev-solucoes/pwdev-claude-marketplace/flow-fleet.svg" alt="Reviewed on agentmods" width="80" height="20"></a>

Per session 59 Skills are progressive disclosure: only the name and description are preloaded; the body loads when the skill is used.

When invoked 910 The whole file, excluding the scripts and references it only reads on demand.

Security scan A 0 findings. A grade says what 26 rules found in the file — not that it is safe.

Origin original No closer match found in the catalogue.

Token cost

What it costs to keep this loaded

Counted locally with the o200k_base tokenizer, which is exact for GPT models; Claude uses its own tokenizer and its counts differ. Treat this as one consistent yardstick across the catalogue rather than a bill. Prices are per million input tokens.

Model	Per session	Once invoked
Fable 5.1	$0.00059	$0.00910
Opus 5	$0.00030	$0.00455
Sonnet 5	$0.00012	$0.00182
Haiku 4.5	$0.00006	$0.00091

Measured 6d ago against content hash f1eca8a82708, method: parsed. Prices are Anthropic first-party input rates as of 2026-09-10, from the pricing page.

Security

Grade A, and why

flow-fleet scanned grade A with 0 findings against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured 6d ago.

A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.

Nothing flagged

None of the 26 patterns this scan looks for appear in this file: no shell pipes, no recursive deletes, no credential paths, no hidden text, no instruction-override or anti-refusal phrasing, no agent-config snooping. That is not a guarantee, it is the absence of the things that are checkable.

plugins/pwdev-flow/skills/flow-fleet/SKILL.md · 48 lines

How it starts

The opening of the file, as written. The whole thing — 48 lines — stays where its author put it; the contents beside it link to each section on GitHub.

Operate a Flow Fleet

Read fleet, safety, artifacts, and collaboration before acting.

Route

Interpret only these routes:

<slugs>: launch one or more approved lowercase phase slugs.
--status: invoke packaged fleet-dashboard.sh --once.
--teardown <slug>: invoke packaged fleet-teardown.sh <slug> and preserve its branch and worktree.
--teardown --all: enumerate validated .planning/flow/fleet/<slug>.json entries and invoke packaged teardown once per slug.

Pass --merge to teardown only when the user explicitly requests it. Never merge a non-DONE member or translate state into legacy paths.

Launch

Inspect repository status, confirm a named current branch, inspect .planning/flow/config.json, and inspect the requested approved spec.md and decisions.md contracts. Never read .env.fleet.
Resolve scripts from this installed skill at ../../scripts/; call only packaged fleet and audit scripts for lifecycle operations.
Use the launcher of the runtime you are running as, and never the other one. The runtime is fixed by that choice, is recorded in the central member, and is refused by a runner bound to a different one. Before the first launch, display the exact autonomous command shape of that runtime:

Codex — fleet-up.sh:
```
codex exec --dangerously-bypass-approvals-and-sandbox --ephemeral --cd <worktree> --output-schema <fleet-result-schema> --output-last-message <result-file> <autonomous-prompt>
```
Claude — claude-fleet-up.sh:
```
claude -p --dangerously-skip-permissions --no-session-persistence --output-format json <autonomous-prompt>
```
Require explicit acknowledgement of the dangerous flag of that runtime — --dangerously-bypass-approvals-and-sandbox for Codex, --dangerously-skip-permissions for Claude. Do not persist a missing fleet block or launch a member before acknowledgement. Reject a configured permission_mode other than danger-full-access.
After acknowledgement, require the configuration root and any existing .fleet value to be objects. Reject an existing permission_mode unless it is exactly danger-full-access; require the exact JSON boolean auto_simplify: false and exact packaged Compose filename before launch.
Merge the exact defaults from fleet into both absent and partial fleet blocks with jq: set .fleet = ($fleet_defaults * (.fleet // {})). Write through a same-directory temporary file and atomically rename it only after validation. This defaults-first recursive merge fills missing known fields while every existing known or unknown field wins; never overwrite permission mode to dangerous.
For multiple slugs, compare their specification and decision text for repeated repository paths. Emit an advisory overlap warning with the matching paths; let the human decide and never block automatically.
Invoke the packaged launcher of your runtime once per acknowledged slug — fleet-up.sh <slug> as Codex, claude-fleet-up.sh <slug> as Claude. Either one binds exact contract hashes and initiating branch/base identity, records the runtime, and rejects unsafe symlinked state paths. Both route their tmux pane to the same autonomous runner, which owns locks, process-group ownership, structured result validation, per-stage commits, and the two-cycle correction cap. Stop launching further members after any failure and report the recovery state.

Read the full file on GitHub · 48 lines

Files

What ships with it

1 file beside SKILL.md in the same directory: the scripts, references and assets a skill reads on demand. Not counted in the per-session cost; read them before you install if any of them is executable.

agents/openai.yaml 191 B

Changes

What this file has done since we first saw it

Hashed on every crawl. A supply-chain change to an agent config is a question of when, not whether, so the history is kept rather than the latest state alone.

6d ago First seen · 48 lines · 59 tokens per session scan A f1eca8a82708

Subscribe to this mod's changes

flow-fleet is a skill published in the GitHub repository pwdev-solucoes/pwdev-claude-marketplace (2 stars, last pushed 3d ago), licensed Apache-2.0. It adds 59 tokens to every session and 910 once invoked, about $0.0003 per session on Opus 5. A static security scan graded it A with 0 findings. No closer match exists in the catalogue, so it is treated as the original; first seen 2026-09-03.

Git and workflow Project management

mcore-split-pr

Split a PR into multiple PRs to reduce the number of required CODEOWNERS reviewer groups.

NVIDIA/skills · 25 tokens

shiplog

Recap of everything shipped since the last run - cross-repo PRs, security fixes, star deltas, and X traction, synthesized into a digest article and a ready-to-post shiplog in your voice.

aeonfun/aeon · 46 tokens

llxprt-issue-workflow

Use this skill when asked to address, fix, or work on a GitHub issue in the llxprt-code repository. Covers the complete issue lifecycle - branch setup, gh CLI usage, test-first planning, subagent delegation and review loops, the full verification cycle (including the stepfun-37 smoke test), open code review (ocr), PR…

vybestack/llxprt-code · 87 tokens

flow-fleet