release

A release-preparation workflow for the ragforgit project, including Russian release notes, version changes, and rebuilt plugin descriptions.

In plain words
What is it for?
Use it when publishing a new version, updating CHANGELOG.md, increasing the version number, or preparing a development-branch merge into the main branch.
Why use it?
It gathers the actual user-facing changes instead of relying only on short commit titles. It also checks whether there is anything new to release.

Skill for Claude CodeCodex

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 skills/mimfort/rag_for_git/release
Any agent
npx skills add mimfort/rag_for_git --skill release
Clone the repo
git clone --depth 1 https://github.com/mimfort/rag_for_git

Made for: Claude Code, Codex.

Per session 76 Skills are progressive disclosure: only the name and description are preloaded; the body loads when the skill is used.
When invoked 1,662 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 $0.00076 $0.01662
Opus 5 $0.00038 $0.00831
Sonnet 5 $0.00015 $0.00332
Haiku 4.5 $0.00008 $0.00166

Measured 2d ago against content hash 4088bb7074a4, method: parsed. Prices are Anthropic first-party input rates as of 2026-08-30, from the pricing page.

Security

Grade A, and why

release 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.

.claude/skills/release/SKILL.md · 127 lines

How it starts

The opening of the file, as written. The whole thing — 127 lines — stays where its author put it; the contents beside it link to each section on GitHub.

Release

Reply in Russian. This skill is maintainer-only tooling for this repository — it is deliberately NOT part of plugin/, so it never ships to plugin users and never changes codex manifests digests.

Run it on dev, before opening the dev → main PR. CI does the rest: on push to main .github/workflows/publish.yml publishes to PyPI, then the release job cuts tag vX.Y.Z and creates a GitHub Release whose body is the CHANGELOG.md section for that version.

1. Determine the range

git fetch --tags origin
git describe --tags --abbrev=0 --match 'v*'   # last released version
git log --no-merges --pretty='%h %s' v<last>..HEAD
git diff --stat v<last>..HEAD

If there are no commits since the last tag, say so and stop — there is nothing to release.

2. Gather what actually changed

Commit subjects alone are NOT enough. For each cluster of related commits, find out what the user gains or stops suffering from:

  • gh pr list --state merged --base dev --limit 30 --json number,title,body,mergedAt — PR bodies usually carry the "why".
  • Task keys in commit messages (PRI-NNN) — read the task via get_task if the reviewer MCP server is available.
  • docs/superpowers/specs/ and docs/plans/ — briefs written when the work started.
  • Read the diff of anything you cannot explain. Never guess at a change's purpose.

3. Decide the version

Semver against the previous release: breaking change to a CLI flag, MCP tool contract, .review.yml schema or stored data → minor while pre-1.0 (and call it out); new capability → minor; fixes and internals only → patch. Propose the number and let the user override it.

4. Write the notes

Into CHANGELOG.md, directly under the intro block, in Russian, following the existing sections:

## X.Y.Z — YYYY-MM-DD

### Что нового
### Изменено
### Исправлено
### Ломающие изменения

Omit empty rubrics. Rules that make the notes worth reading:

  • Group by meaning, not by commit — but grouping means one heading per feature, not one sentence. Fifteen commits that built one feature become one bullet with sub-bullets, never a single line that mentions the feature exists.
  • Never swallow an enumerable surface. If the change introduces named things the user will choose between or type — modes, strategies, flags, commands, config keys, statuses — every name goes into the notes, each with what it does and when to pick it. "Added a start-up prompt for mode and strategy" is a failed note: the reader still does not know that full-auto and lite exist. This is the single most common way these notes go wrong; check for it before publishing.
  • Lead with the capability, not the module. "Поднять инфраструктуру одной командой" beats "feat(cli): команды reviewer start и reviewer stop".
  • Say what it changes for the reader. A fix names the symptom that disappears; a feature names the thing that was previously impossible or manual. State the cost or caveat where one exists — when a mode is a bad idea, say so in the note rather than letting the reader discover it.
  • Read the spec, not just the commits. A feature built from a docs/superpowers/specs/ design has its full surface listed there; commit subjects show only the construction order. If a spec exists and you did not open it, you are guessing at what shipped.
  • Skip pure noise — manifest rebuilds, uv.lock, version bumps, internal test fixups — unless it changes something observable.
  • Breaking changes get their own rubric with the exact migration step.
  • Length follows the release: one bullet per user-visible change, sub-bullets for its named parts. A release with three features and a rich one among them runs long, and that is correct — the budget is "no filler", not "no detail". If there is genuinely little to say, say it plainly instead of padding.

Read the full file on GitHub · 127 lines

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 · 127 lines · 76 tokens per session scan A 4088bb7074a4

Subscribe to this mod's changes

release is a skill published in the GitHub repository mimfort/rag_for_git (21 stars, last pushed 2d ago), licensed MIT. It adds 76 tokens to every session and 1,662 once invoked, about $0.0004 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 2026-08-30.

Related

Other skills, from other repositories