Skip to content
Development
Skill

/rig-release

Assess release readiness with a version baseline from HEAD, full gates, packaged file verification, and read-only release-note, migration, and contributor-reply drafts.

BOOST
From plugin
ai-rig
2948 skills16 agents2 MCP
Install
$ npx -y skills add Borda/AI-Rig --skill rig-release --agent claude-code

How it fires

How this skill gets triggered: by you, by Claude, or both.

  • Fires itselfAuto-invocation. Claude auto-loads it when your prompt matches the work.Auto-invocation is when the right skill fires by itself at the right moment, driven by a FLOW.md router and a hook, instead of you invoking it by name. It is the difference between a skill being installed and a skill actually getting used.Read the full definition →
  • You can call itInvoke it directly when you want it.
  • Slash command/rig-release

Context preview

The summary Claude sees to decide when to auto-load this skill.

Assess release readiness with a version baseline from HEAD, full gates, packaged file verification, and read-only release-note, migration, and contributor-reply drafts.

SKILL.md

rig-release.SKILL.md
name: Rig Release
description: Assess release readiness with a version baseline from HEAD, full gates, packaged file verification, and read-only release-note, migration, and contributor-reply drafts.

Rig Release

> Read [Evidence rules](../../references/evidence.md).

> This workflow never publishes, tags, pushes, or mutates a remote. Publication stays human-owned.

Workflow

1. **Read the baseline from HEAD.** Do not use an uncommitted version. For this plugin family, `package.json` is the version manifest; confirm the repository's policy for the package being released. 2. **Classify the change.** Fix or wording: patch. New capability or surface: minor. Breaking contract: major, and only with explicit approval. 3. **Start the run.** `rig_evidence` with `action: "start"`, `skill: "rig-release"`, and the package root. 4. **Run the full gate set.** `rig_gate` with `action: "run"` and no name filter. A single failure blocks readiness; do not retry unchanged. 5. **Synchronize documentation.** Verify the changelog bullet describes user-visible impact, the README version marker matches the manifest, and changed agents, skills, tools, and commands are documented. 6. **Draft the release pack.** Bind the commit range with `rig_evidence` `action: "record"` and `commit`, then produce the three drafts below from that evidence. Every line must trace to a commit, finding, or changelog entry; an empty section is better than an invented one. 7. **Verify the package.** Run the package build or `npm pack --dry-run` (or the project equivalent) and check the file list. No tests, plans, credentials, or private evidence may ship. 8. **Compare versions.** Declared version must equal `HEAD` plus exactly one SemVer step unless the package is new. Record the comparison. 9. **Finalize.** Produce a readiness verdict with blockers, the drafts, and a human checklist.

Release pack

  • **Release notes.** Group by Added, Changed, Fixed, Security, Deprecated, and Removed. Each entry names the user-visible effect, not the internal refactor, and credits the contributor when the commit or issue records one.
  • **Migration notes.** List breaking changes, renamed options, changed defaults, and required user actions, each with before/after snippets and a rollback path. Say explicitly when no action is required.
  • **Contributor replies.** Draft a short factual reply per linked issue or pull request: what shipped, what did not, and what happens next. Never promise a date, never claim a fix that is not in the range, and never post it.

Rules

  • A dirty working tree is not a blocker for assessment, but it must be disclosed; the released commit must be clean.
  • Never bump a version skips a step or down from `HEAD`.
  • Do not publish, tag, push, comment, or open a pull request, even when every check passes. Drafts stay in the report.
  • Never invent a changelog entry, contributor, issue link, or migration step. Missing evidence is a blocker, not a blank to fill.
  • Report unavailable evidence as a blocker, not a pass.

Output

Readiness verdict, proposed version, gate results, documentation state, package file list, release notes, migration notes, contributor reply drafts, blockers, and the remaining human steps.

Read more
Ships withai-rig

Practical agent workflows for Python, ML, and open-source maintenance. AI-Rig turns recurring work—scoping a change, reproducing a bug, reviewing a pull request, running an experiment, or checking release readiness—into explicit workflows with specialist

Get the whole plugin

Other skills on ai-rig.

debug
Skill

debug

Investigation-first debugging — gather evidence, form confirmed root-cause hypothesis, hand off to fix mode with diagnosis file.…

@borda@bordaInstall Skill
feature
Skill

feature

TDD-first feature development — crystallise API as a demo test, drive implementation to pass it, run quality stack and…

@borda@bordaInstall Skill
fix
Skill

fix

Reproduce-first bug resolution — capture bug in failing regression test, apply minimal fix, run quality stack and review loop.…

@borda@bordaInstall Skill
plan
Skill

plan

Analysis-only planning — classify and scope a task without writing code; outputs a structured plan to .plans/active/. TRIGGER…

@borda@bordaInstall Skill
refactor
Skill

refactor

Test-first refactoring — audit coverage, add characterization tests, apply changes with safety net, run quality stack and review…

@borda@bordaInstall Skill
review
Skill

review

Multi-agent code review of local Python files, directories, or the current git diff covering architecture, tests, performance,…

@borda@bordaInstall Skill