Agent skill

Mecatl Release Cutting

by stacklok in stacklok/mecatl

Cuts a tagged mecatl release by dispatching the release-PR workflow, merging the bot's pull request and verifying the tag, images, Helm chart, signed archives and Homebrew formula.

Apache-2.0Auto-check passedDevOps & Cloud

Install Mecatl Release Cutting

skills CLI
$ npx skills add stacklok/mecatl --skill cut-release -a claude-code

Project install by default; add -g for ~/.claude/skills/.

GitHub CLI
$ gh skill install stacklok/mecatl cut-release --agent claude-code

Project scope by default; add --scope user for a personal install. Needs GitHub CLI 2.90.0 or later (public preview).

Manual copy
$ git clone --depth 1 https://github.com/stacklok/mecatl.git skills-src && mkdir -p .claude/skills && cp -r skills-src/.claude/skills/cut-release .claude/skills/cut-release && rm -rf skills-src

Use ~/.claude/skills/ instead of .claude/skills for a personal install. The folder must contain SKILL.md.

Claude Code skills documentation · loads skills from .claude/skills/

Facts

Skill name
cut-release
GitHub stars
218
Token cost
~4k tokens
SKILL.md length
2,050 words
Files
1
Skills in repo
7
Repo updated
First seen
Licence
Apache-2.0

At a glance

Cuts a tagged mecatl release by dispatching the release-PR workflow, merging the bot's pull request and verifying the tag, images, Helm chart, signed archives and Homebrew formula.

  • Works in 8 steps: Confirm what you're shipping. The… → Dispatch the release-PR workflow. This… → Review the release PR like any other PR… → …
  • Cutting or shipping a new mecatl release
  • SKILL.md covers Steps, The engine module is tagged… and Notes
  • Calls gh, git and brew; needs GITHUB_TOKEN and RELEASE_APP_PRIVATE_KEY

What it does

A mecatl release is a version tag. Pushing it triggers the release workflow, which publishes signed container images, Helm charts, microVM and Brood Box artifacts, a GitHub Release with darwin and linux CLI archives for amd64 and arm64, checksums, cosign bundles, SBOMs and build provenance, and a formula bump in the public Homebrew tap. No version is baked into the Go code. Nothing pushes to `main` directly: you dispatch a workflow, a bot opens the release PR, a human merges it and a bot tags the merge commit, so the agent never pushes main or creates the tag by hand.

The root `VERSION` file, in bare semver form, is the source of truth and the release PR propagates it to the chart version, app version and default image tag. A published tag is treated as immutable because the Homebrew formula records archive checksums, so a bad release after the tap commit is fixed forward with the next patch version, and re-tagging is allowed only if the run failed before publishing. Release-only inputs are validated before tagging, not with a throwaway tag. The excerpt is cut off in a section about composite action pins.

When your agent uses it

  • Cutting or shipping a new mecatl release
  • Verifying the tag, images and Homebrew bump after a release run
  • Deciding how to recover from a release that failed after publishing

Example prompts

  • “Cut the next patch release of mecatl and verify the published artifacts.”
  • “Dispatch the Create Release PR workflow and tell me when it is ready to merge.”
  • “The release run failed after the tap commit landed, so what should we do?”

Requirements

  • The mecatl repository and its release workflows
  • GitHub access to dispatch workflows and merge pull requests

Workflow steps

8 steps, taken from the first numbered list in SKILL.md.

  1. Confirm what you're shipping. The release tags whatever is on main when the release PR
  2. Dispatch the release-PR workflow. This is the only step that starts a release
  3. Review the release PR like any other PR and confirm that all changes belong to the
  4. Squash-merge it. A human does this — it is the approval gate, and it is the only way
  5. Watch the tag get created. Merging fires create-release-tag.yml, which re-verifies the
  6. Confirm the release run started, then wait for every publishing job. A
  7. Verify the GitHub Release carries every artifact. It must not be a draft, and it must
  8. Verify the Homebrew tap got the formula bump

What it can do on your machine

Read from SKILL.md and the folder at commit e731897. It shows what the files ask for, not the result of running them.

  • Tool permissions

    Pre-approves nothing: there is no allowed-tools line, so your agent's usual permission prompts apply.

    From allowed-tools in the SKILL.md frontmatter.

  • Runs code

    Shell commands in SKILL.md call:

    • gh
    • git
    • brew
    • go

    From the folder's file list and the shell code blocks in SKILL.md.

  • Network

    No URLs in SKILL.md. Its commands use gh and git, which can reach the network depending on how they are called.

    From URLs in SKILL.md, links to its own repository left out.

  • Credentials

    Names these keys or tokens, usually read from environment variables:

    • GITHUB_TOKEN
    • RELEASE_APP_PRIVATE_KEY

    From names ending in _API_KEY, _TOKEN, _SECRET, _KEY or _PASSWORD in SKILL.md.

Context cost

Mecatl Release Cutting loads about 4k tokens when it runs. Until then it costs about 101 tokens; SKILL.md has 2,050 words of instructions outside code blocks.

Always · name and description, kept in context so the agent knows when to use it
~101
When it runs · the whole SKILL.md, loaded when a task matches
~4k

Estimates: characters ÷ 4, the usual rule of thumb; real counts depend on the model's tokenizer. Scripts and assets cost tokens only if the agent reads them.

Safety

Auto-check passed

The automated check found no risky patterns in SKILL.md.

Automated static check — not a guarantee. Review scripts before installing. It scans the text of SKILL.md for risky patterns (piping downloads into a shell, reading credential files, hidden Unicode, destructive commands); files beside SKILL.md are not scanned.

SKILL.md

The full file from stacklok/mecatl at commit e731897, republished under its Apache-2.0 licence (© stacklok). 2,050 words, ~4,013 tokens.

Download SKILL.mdSave it as .claude/skills/cut-release/SKILL.md (or your agent's skills folder).
name
cut-release
description
Cut a tagged release of mecatl — dispatch the Create Release PR workflow, review and merge the release PR, then verify the tag and the artifacts it publishes (ko images + Helm chart to GHCR, plus a GitHub Release with signed archives and a Homebrew formula bump). Use when asked to cut/ship/tag/publish a release or bump the version. NOT for general git tagging unrelated to a mecatl release.
metadata.author
stacklok

Cut a mecatl release

A release is a vX.Y.Z git tag. Pushing that tag triggers .github/workflows/release.yml, which publishes signed images (mecated, mecatui, mecak8s, execution provider/workload, Studio, and the Slack-bot example), Helm charts, and versioned microVM and Brood Box artifacts. It also publishes a GitHub Release carrying darwin/linux x amd64/arm64 CLI archives, a checksums.txt, cosign bundles, SBOMs, and build provenance, and pushes a mecatl formula bump to the public stacklok/homebrew-tap repository. There is no version baked into the Go code — the tag IS the release.

Two consequences of the Homebrew half, before you start:

  • A published tag is immutable in practice. The formula records the release archives' SHA256 checksums for a specific tag. Deleting and re-pushing a vX.Y.Z that already produced a release and a tap commit leaves the tap pointing at checksums that no longer match, which breaks brew install for everyone. If a release goes wrong after the tap commit lands, fix forward with the next patch version. Re-tagging is only an option when the run failed before publishing anything.
  • Validate release-only inputs BEFORE tagging, not by pushing a throwaway tag. The release-PR workflow checks execution-image digest pins before it opens the PR; CI checks the Brood Box build and microVM defaults handoff. A local task release:snapshot && task release:verify only checks CLI archives and Homebrew packaging; it does not prove the container publishing jobs work.

Nothing pushes a commit to main. The release runs through an ordinary pull request: you dispatch a workflow, a bot opens the PR, a human merges it, and a bot tags the merge commit. You never run git push origin main, and you never create the tag by hand.

VERSION (repo root, bare semver — 0.0.34, not v0.0.34) is the source of truth for the release version. The release PR propagates it to the mecak8s chart version, appVersion, and default image tag. The workflows verify those values directly rather than maintaining a fixed list of files that a release PR may change.

It did not used to be. mecatequi-reusable.yml referenced its three sibling composite actions by a hardcoded @vX.Y.Z literal, so every release had to bump those pins in the same tagged commit or ship version skew. That self-reference — a file naming a tag that does not exist yet — is why a release needed a commit on main at all. The pins are now $/ self-repository refs, which resolve to this repo at the exact ref the workflow is running from, so there is nothing left to bump and skew is impossible rather than merely policed.

Steps

Run from the repo root.

  1. Confirm what you're shipping. The release tags whatever is on main when the release PR merges. Review what has landed since the last tag:

    sh
    git tag --sort=-v:refname --list 'v*' | head -1   # e.g. v0.0.33
    git log <last-tag>..origin/main --oneline

    Confirm the bump type with the operator before dispatch; patch is the established release cadence even when the range includes additive changes. Do not infer a minor bump from commit subjects alone.

  2. Dispatch the release-PR workflow. This is the only step that starts a release:

    sh
    gh workflow run create-release-pr.yml -f bump_type=patch
    gh run watch "$(gh run list --workflow=create-release-pr.yml --limit 1 --json databaseId --jq '.[0].databaseId')"

    The workflow validates the tracked ARG defaults in build/execution-provider/Dockerfile and build/execution-workload/Dockerfile on main: the Go builder, static provider runtime, and Brood Box workload runtime must have tagged digest pins available for Linux amd64 and arm64; it also builds both images without publishing. The tag workflow repeats the pin and availability validation on the merged commit before tagging. Renovate updates the pins through its native Dockerfile manager; no image repository variables are required. If validation fails, fix the tracked pins through review (or retry a transient registry failure); do not bypass the check. The publishing job validates the tagged checkout again. The workflow bumps VERSION, the mecak8s chart version and app version, and the chart's default image tag. It opens Release vX.Y.Z from branch release/vX.Y.Z, then asserts that the required values are synchronized. If that verification step fails, do not merge the PR; close it, delete the branch, and read the job log.

  3. Review the release PR like any other PR and confirm that all changes belong to the release update:

    sh
    gh pr list --head "release/vX.Y.Z" --json number,url,files
    gh pr diff <number>

    Wait for CI to go green. The PR is opened by the release GitHub App, so it triggers checks normally.

  4. Squash-merge it. A human does this — it is the approval gate, and it is the only way VERSION changes on main:

    sh
    gh pr merge <number> --squash

    The tagging workflow does not read the commit subject — it asks GitHub which PR produced the commit and requires a merged, bot-opened PR from branch release/vX.Y.Z with matching version values. The squash title and the number of updated files do not affect the tag gate.

  5. Watch the tag get created. Merging fires create-release-tag.yml, which re-verifies the commit and pushes the annotated tag. That push fires release.yml on its own — the tag is pushed by a GitHub App installation token precisely so the cascade happens, where a GITHUB_TOKEN-pushed tag would trigger nothing:

    sh
    gh run watch "$(gh run list --workflow=create-release-tag.yml --limit 1 --json databaseId --jq '.[0].databaseId')"
    git fetch --tags && git tag --sort=-v:refname --list 'v*' | head -1
  6. Confirm the release run started, then wait for every publishing job. A successful CLI archive or one green image is not a complete release. In particular, check the mecated, execution provider/workload, Brood Box resolution, microVM, chart, and CLI/Homebrew jobs before reporting success:

    sh
    gh run list --workflow=release.yml --limit 3
    gh run watch "$(gh run list --workflow=release.yml --limit 1 --json databaseId --jq '.[0].databaseId')"
  7. Verify the GitHub Release carries every artifact. It must not be a draft, and it must have four archives plus a checksum file, with a cosign bundle and an SBOM alongside each. Also require brood-base-index.json, brood-platforms.json, and the microvm-default-*.json completion assets for each platform. Missing assets mean the release is incomplete even if the CLI archives are available:

    sh
    gh release view vX.Y.Z --json isDraft,assets --jq '{draft: .isDraft, assets: [.assets[].name]}'

    Then prove one archive is actually usable rather than trusting the asset list. Use the repo-local .scratch/ dir, never /tmp (AGENTS.md):

    sh
    mkdir -p .scratch/release-vX.Y.Z
    gh release download vX.Y.Z -p 'checksums.txt' -p '*darwin_arm64*' -D .scratch/release-vX.Y.Z
    (cd .scratch/release-vX.Y.Z \
      && shasum -a 256 -c checksums.txt --ignore-missing \
      && tar -xzf mecatl_*_darwin_arm64.tar.gz \
      && ./mecatui --version && ./mecated --version)

    Both --version lines must print the tag you just cut. A dev+<revision> output means the release build lost its BUILD_ID linker stamp — a release bug, not a cosmetic one, because the public install docs claim a released binary reports its tag.

    If the run died between "release created" and "assets uploaded" it leaves a DRAFT, which a lookup by tag does not return, so a naive re-run fails trying to create the release again. Recover with gh release delete vX.Y.Z --cleanup-tag=false --yes, then re-dispatch.

  8. Verify the Homebrew tap got the formula bump:

    sh
    gh api repos/stacklok/homebrew-tap/commits --jq '.[0].commit.message'
    gh api repos/stacklok/homebrew-tap/contents/Formula/mecatl.rb --jq '.content' \
      | base64 -d | grep -E 'version|url|sha256' | head

    The top commit must name the version you just cut, and the formula's url and sha256 values must match the release assets from step 7.

    While stacklok/mecatl is private, brew install stacklok/tap/mecatl fails even after a correct tap commit: Homebrew's downloader does not authenticate, so it cannot fetch a release archive from a private repository. The tap commit landing is the whole verification until the repository goes public; this is known and accepted. Once it is public, run the real end-to-end check once:

    sh
    brew update && brew install stacklok/tap/mecatl && mecatui --version

The engine module is tagged separately

The vX.Y.Z release above is the root repo / mecated image release. The importable core, github.com/stacklok/mecatl/engine, is its own Go module (ADR 0036) with its own tag grammar engine/vX.Y.Z (distinct from the root tags). It carries a public-API compatibility contract (engine/COMPATIBILITY.md, ADR 0037).

  • The first engine/vX.Y.Z tag is engine/v0.0.1 — a deliberate "earliest, no stability promise" initial cut (the lowest pre-v1 patch, signalling zero stability commitment for the very first published surface). Cutting it is a deliberate maintainer decision (deferred per ADR 0037) — do NOT cut it as part of a routine root release unless asked. The grammar is engine/vX.Y.Z, distinct from the root vX.Y.Z tags; the two version lines are independent. SUBSEQUENT bumps follow engine/COMPATIBILITY.md (pre-v1: minor = additive, patch = fixes).
Show full SKILL.md (849 more words)Show less
Cutting an engine tag (mirrors the root flow)

Run from the repo root.

  1. Pick the engine version. First cut = engine/v0.0.1 (a deliberate "earliest, no stability promise" initial cut); thereafter increment per semver, classified per engine/COMPATIBILITY.md (pre-v1: Added = minor, Changed/Removed = minor too; patch = fixes). The latest engine tag (none yet on the first cut):

    sh
    git tag --sort=-v:refname --list 'engine/v*' | head -1
  2. Pre-flight. Confirm engine/CHANGELOG.md has an [Unreleased] entry covering everything since the last engine tag (on the first cut that is the whole initial surface — the existing [Unreleased] baseline section). Then run the advisory gorelease check:

    sh
    task api:release-check

    On the FIRST cut this is a no-op / uninformative: gorelease can only classify the surface against a prior engine/vX.Y.Z base tag, and none exists yet — so it has nothing to compare to. That is expected. The authoritative guard is the api-compat gate (task api:check), which already guarantees the committed engine/api/*.txt snapshots match the surface being tagged.

  3. Create the annotated tag with a concise summary:

    sh
    git tag -a engine/vX.Y.Z -m "engine/vX.Y.Z — <one-line summary>"
  4. Push the tag:

    sh
    git push origin engine/vX.Y.Z

This line is deliberately still manual. An engine tag adds no commit to main and carries no pin bump, so it never needed the release-PR flow the root vX.Y.Z line uses — pushing the tag is the whole release.

IMPORTANT — an engine tag fires NO image build, NO GitHub Release, and NO Homebrew formula bump. release.yml triggers on v* (the root tag glob), which does not match engine/v*, so cutting an engine tag runs none of the ko build / cosign / SBOM / SLSA pipeline. It only publishes the module version, making it resolvable for go get github.com/stacklok/mecatl/engine@engine/vX.Y.Z consumers (ADR 0036/0037). There is no pin bump and no release.yml run to confirm — the push of the tag is the whole release.

Notes

  • Two publishing destinations, one tag. A run can succeed on the GHCR images and still fail on the release or the tap (or vice versa). GoReleaser's brew pipe continues on error and the Release is created before the formula is pushed, so a bad tap token loses the formula but NOT the Release. Steps 7 and 8 are not optional: a green gh run list line is not proof that brew install works.
  • Never hand-edit stacklok/homebrew-tap. The formula is generated from the tag by the release workflow and carries a DO NOT EDIT header. A manual edit is overwritten by the next release and desynchronizes the checksums in the meantime.
  • Never push to main, and never create a root vX.Y.Z tag by hand. Both are the workflows' job. A hand-pushed version bump skips code review, and a hand-created tag can point at a commit whose release metadata the gate rejects. If VERSION is edited on main outside a release PR, create-release-tag.yml refuses to tag it rather than cutting a release from it. release.yml's guard job additionally refuses to publish anything from a tag that is not an ancestor of main, so a tag cut on a branch builds nothing. (This applies to the ROOT v* line only — the engine/v* tags below are still cut by hand, deliberately: they carry no pin bump and add no commit to main.)
  • Annotated tags only (git tag -a), matching prior releases — they carry a tagger + message. create-release-tag.yml does this; the tagger is github-actions[bot].
  • Don't bump illustrative documentation refs unless asked — the @vX.Y.Z examples in user-docs/building/deployment/mecatequi.md are illustrative and do NOT gate the release. The release flow deliberately leaves them alone.
  • If the release run fails on the version gate, the tagged commit has inconsistent release metadata. That should be impossible through the normal flow: create-release-pr.yml verifies the bump before the PR can merge, and create-release-tag.yml tags only the merge commit. It means someone tagged by hand or the chart metadata drifted from VERSION. Fix forward with a patch.
  • Rerunning is safe. create-release-tag.yml makes one decision from the tag's state and the commit's provenance, so it is quiet when there is nothing to do (the tag already points here, or VERSION names an already-released tag this commit did not produce) and loud only when a tag should have been created and something is wrong. release.yml re-signs idempotently via its workflow_dispatch tag input.
  • Setup, once. Both workflows read the release GitHub App from a release GitHub Environment (vars.RELEASE_APP_CLIENT_ID, secrets.RELEASE_APP_PRIVATE_KEY), whose deployment-branch policy must be restricted to main. The App needs exactly two repository permissions — Contents: write and Pull requests: write. It does NOT need Workflows: write, because a release no longer edits anything under .github/workflows/. Repo-level secrets would let anyone with push access dispatch a modified workflow from a branch and mint the App credential.
  • One release at a time. If any release/v* PR is open, the next dispatch refuses and names it — merge or close it first. Once none is open, leftover release/v* branches from failed runs are deleted automatically before the new PR is cut.
  • The release gate checks synchronized values, not a fixed file list. This lets the release automation add another version projection or generated file without requiring a second gate update. There is nothing to dry-run locally.

© stacklok, Apache-2.0. Rendered from Markdown: HTML in the file is shown as text, images as links, and headings moved down two levels. Raw file

Files

Just SKILL.md in .claude/skills/cut-release of stacklok/mecatl.

Open the folder on GitHubat commit e731897

Compare with similar skills

Mecatl Release Cutting next to the 5 skills that share the most tags, products or categories with it. Stars are the repository's; “used in” counts other GitHub owners with a copy.

Mecatl Release Cutting compared with similar skills
SkillStarsUsed inTokensAuto-checkLicenceRepo updated
Mecatl Release Cutting this skillstacklok/mecatl218—~4kAutomated safety check: PassApache-2.0
ClickUp CLI Release Processkrodak/clickup-cli120—~906Automated safety check: WarnMIT
OpenWork Release Processdifferent-ai/openwork24k—~2.3kAutomated safety check: PassCustom licence
ccLoad Release Publishercaidaoli/ccLoad417—~887Automated safety check: PassMIT
AnyDrag Release RoutineXueshiQiao/AnyDrag227—~2.6kAutomated safety check: PassGPL-3.0
Kt Search Releasejillesvangurp/kt-search155—~1.2kAutomated safety check: PassMIT

Similar skills

  • ClickUp CLI Release Process

    krodak/clickup-cli

    Walks through releasing a new version of clickup-cli: pre-release checks, version bump, tagging, CI watch, release notes and the Homebrew update.

    120 GitHub stars~906 tokensUpdated today
    DevOps & CloudAuto-check: warnings
  • OpenWork Release Process

    different-ai/openwork

    Cuts an OpenWork desktop release through a tag-driven GitHub Actions workflow that makes no commits, with pre-tag checks on open fix PRs and verification afterward.

    24k GitHub stars~2.3k tokensUpdated today
    DevOps & CloudAuto-check passed
  • Publishes a ccLoad Beta or explicit stable release through a version-tag workflow, including commit, push, CI wait, GitHub Release and container image checks.

    417 GitHub stars~887 tokensUpdated today
    DevOps & CloudAuto-check passed
  • AnyDrag Release Routine

    XueshiQiao/AnyDrag

    Runs the full AnyDrag release process end to end, from cumulative bilingual release notes through version bumping to watching CI and the Homebrew cask update.

    227 GitHub stars~2.6k tokensUpdated 2 days ago
    DevelopmentAuto-check passed
  • Kt Search Release

    jillesvangurp/kt-search

    A skill your agent uses when the user wants to cut, publish, tag, or create a GitHub release for kt-search, especially when the task includes version bumping, validating that commits are pushed…

    155 GitHub stars~1.2k tokensUpdated 3 mo ago
    DevelopmentAuto-check passed
  • Agr Release

    computerlovetech/agr

    Release process for the agr package. An agent skill from computerlovetech/agr.

    451 GitHub stars~1.6k tokensUpdated 1 mo ago
    DevOps & CloudAuto-check passed

More from stacklok/mecatl

  • Interviews you about provider, cost, openness and image needs, then designs the models section of a mecatl settings file with aliases, slots and router categories.

    218 GitHub stars~2.7k tokensUpdated today
    Auto-check passed
  • Runs mecatl's offline benchmark and scenario harness to measure, profile with pprof, optimize and prove a performance win with benchstat, then adds a regression benchmark.

    218 GitHub stars~2.2k tokensUpdated today
    Auto-check passed
  • mecatl Learning Config

    stacklok/mecatl

    Designs, validates and writes the learning section of a mecatl settings file, covering mode, sensitivity, reflection budgets and validated or evaluated activation.

    218 GitHub stars~3.8k tokensUpdated today
    Auto-check passed
  • Guides reading mecatl's perf MCP data to find why a running harness is slow, leaking goroutines or growing in memory, using cheap reads before any CPU capture.

    218 GitHub stars~2.3k tokensUpdated today
    Auto-check passed
  • Rebuilds the mecak8s image into the local mecatl-dev Kind cluster and builds mecatui, so you can try in-progress mecatl changes against a real Kubernetes deployment.

    218 GitHub stars~927 tokensUpdated today
    Auto-check passed
  • Panel Review

    stacklok/mecatl

    Review completed non-trivial code across four independent axes: Spec, Standards, Test adequacy, and installed Domain specialists.

    218 GitHub stars~5.8k tokensUpdated today
    Auto-check passed

Questions about Mecatl Release Cutting

What does Mecatl Release Cutting do?

Cuts a tagged mecatl release by dispatching the release-PR workflow, merging the bot's pull request and verifying the tag, images, Helm chart, signed archives and Homebrew formula. A mecatl release is a version tag. Pushing it triggers the release workflow, which publishes signed container images, Helm charts, microVM and Brood Box artifacts, a GitHub Release with darwin and linux CLI archives for amd64 and arm64, checksums, cosign bundles, SBOMs and build provenance, and a formula bump in the public Homebrew tap.

When should I use Mecatl Release Cutting?

Mecatl Release Cutting fits situations like: cutting or shipping a new mecatl release; verifying the tag, images and Homebrew bump after a release run; deciding how to recover from a release that failed after publishing.

How do I install Mecatl Release Cutting in Claude Code?

Run `npx skills add stacklok/mecatl --skill cut-release -a claude-code`. Or copy the skill folder (.claude/skills/cut-release in stacklok/mecatl) into .claude/skills/cut-release in your project. Claude Code loads it when a task matches its description.

How do I install Mecatl Release Cutting in Codex?

Run `npx skills add stacklok/mecatl --skill cut-release -a codex`. Or copy the skill folder (.claude/skills/cut-release in stacklok/mecatl) into .agents/skills/cut-release in your project. Codex loads it when a task matches its description.

Can I use Mecatl Release Cutting in Cursor, Gemini CLI or GitHub Copilot?

Cursor, Gemini CLI, GitHub Copilot and OpenCode also load SKILL.md folders. With the skills CLI, run `npx skills add stacklok/mecatl --skill cut-release -a cursor` (or -a gemini-cli, github-copilot or opencode for the others). To copy it by hand, put the folder in .cursor/skills/cut-release, .gemini/skills/cut-release, .github/skills/cut-release and .opencode/skills/cut-release in your project.

What does Mecatl Release Cutting need to run?

Going by SKILL.md and its folder, Mecatl Release Cutting needs the command-line tools its instructions call (gh, git, brew and go) and credentials named GITHUB_TOKEN and RELEASE_APP_PRIVATE_KEY. Our summary lists: The mecatl repository and its release workflows; GitHub access to dispatch workflows and merge pull requests.

Does Mecatl Release Cutting access the network?

SKILL.md contains no URLs. Its commands use gh and git, which can reach the network depending on how they are called. This is read from the text; nothing was executed.

Is Mecatl Release Cutting safe to install?

Our automated static check of SKILL.md found no risky patterns, such as piping downloads into a shell, reading credential files or hidden Unicode. It is not a guarantee. Review the folder before installing.

What licence does Mecatl Release Cutting use?

Mecatl Release Cutting is published under the Apache-2.0 licence (the repository's licence). It allows redistribution, so the full SKILL.md is shown on this page.

How many tokens does Mecatl Release Cutting use?

About 4k tokens (SKILL.md is roughly 16k characters). Agents keep only the skill's name and description in context until a task matches; then they load SKILL.md in full.

What are the alternatives to Mecatl Release Cutting?

Skills that share tags, products or a category with Mecatl Release Cutting: ClickUp CLI Release Process (krodak/clickup-cli, 120 stars), OpenWork Release Process (different-ai/openwork, 24k stars), ccLoad Release Publisher (caidaoli/ccLoad, 417 stars) and AnyDrag Release Routine (XueshiQiao/AnyDrag, 227 stars). The comparison table on this page puts their stars, adoption, token cost, safety result and licence side by side.

Who maintains Mecatl Release Cutting?

stacklok (a GitHub organization) maintains it in stacklok/mecatl, which has 218 GitHub stars. The repository holds 7 skills in this directory. The repository was last updated on October 6, 2026.

Source: stacklok/mecatl on GitHub. Facts on this page come from the repository at the commit we read; the author's words are quoted as theirs.