release-flow

A release process that checks a software change and asks a separately configured agent to review it. A release is accepted only when that review starts with “APPROVED”.

In plain words
What is it for?
Use it before releasing or tagging code, in check mode or through a configured agent such as Claude, Codex, Gemini, or OpenCode.
Why use it?
It adds an independent review and fails closed when the required verdict is missing or negative.

Command

Part of the cool-workflow plugin — 5 skills, 2 commands, 1 agent, 1 MCP server shipped together

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.

agentmods
npx agentmods add commands/coo1white/cool-workflow/release-flow
Clone the repo
git clone --depth 1 https://github.com/coo1white/cool-workflow

Or install cool-workflow, the plugin that ships this one along with the rest of its 5 skills, 2 commands, 1 agent, 1 MCP server.

Per session 20 Only the description is in the session, so the agent can decide to use it. The body loads when it is invoked.
When invoked 591 The whole file, excluding the scripts and references it only reads on demand.
Security scan A 0 findings. Scan, not verified.
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.

ModelPer sessionOnce invoked
Fable 5 $0.00020 $0.00591
Opus 5 $0.00010 $0.00296
Sonnet 5 $0.00004 $0.00118
Haiku 4.5 $0.00002 $0.00059

Measured 2d ago against content hash e1c8228d0839, method: parsed. Prices are Anthropic first-party input rates as of 2026-08-30, from the pricing page.

Security

Grade A, and why

release-flow 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 2d 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/cool-workflow/commands/release-flow.md · 58 lines

What it actually says

Run the cool-workflow portable release flow. This is the multi-platform sibling of /release: the orchestration lives in one zero-dependency script and the independent review is DELEGATED to whatever agent you configure, so the exact same flow runs under Claude, Codex, Gemini, OpenCode, or a plain shell.

Steps:

  1. Configure the reviewer agent for THIS host (Claude):

    export CW_AGENT_COMMAND="claude -p --permission-mode acceptEdits {{input}}"
    export CW_AGENT_MODEL="claude-opus-4-8"   # optional; the agent picks otherwise
    # acceptEdits lets the headless reviewer WRITE its verdict file; without it the
    # review runs but the verdict can't be persisted and the flow fails closed at
    # [3/3]. The reviewer CLI must also be logged in (claude auth login).
    
  2. Run the flow in check mode (gate + independent review, no mutation):

    node plugins/cool-workflow/scripts/release-flow.js --check
    

    It runs scripts/release-gate.js, then spawns the configured agent (argv-style, shell:false — CW holds no key) to review the candidate and write .cw-release/review-<HEAD>.verdict. The flow fails closed unless the verdict begins with APPROVED.

Report the JSON summary from step 2 (verdict, capability) and stop at the first failure. Do not write the verdict file yourself — the delegated reviewer is the only writer, and release-gate.yml re-checks it in CI.

Stop here. Do NOT run release-flow.js --cut yourself, on this host or any other — cutting a tag needs CW_RELEASE_VERDICT_PRIVKEY, the ed25519 signing key that stays with the release operator and must never be read or held by an agent (see AGENTS.md's "Shipping a release" section). Once review is green, tell the user the candidate is ready and that the operator cuts it themselves:

npm run release -- <x.y.z>

(plugins/cool-workflow/scripts/release-oneclick.js, the normal path) or, for manual control on a non-npm host, the lower-level command run by the operator in their own shell:

node plugins/cool-workflow/scripts/release-flow.js --cut --version <x.y.z> --push

See docs/release-tooling.7.md for the per-platform CW_AGENT_COMMAND presets (Codex / Gemini / OpenCode / DeepSeek).

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.

  1. 2d ago First seen · 58 lines · 20 tokens per session scan A e1c8228d0839

Subscribe to this mod's changes

release-flow is a command published in the GitHub repository coo1white/cool-workflow (2 stars, last pushed 6d ago), licensed BSD-2-Clause. It adds 20 tokens to every session and 591 once invoked, about $0.0001 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 2026-08-31.