Skip to content

ericyanpek/agentcore-private-registry-blueprint / skills/publish-skill

publish-skill

Verified

Use when the user wants to publish a new agent skill (a directory containing `pyproject.toml` + `src/<pkg>/skill_files/SKILL.md`) to the company's private AWS Agent Registry + CodeArtifact. Walks build → CodeArtifact upload → CreateRegistryRecord → optional submit-for-approval. Requires AWS credentials with codeartifact:PublishPackageVersion and agent-registry:CreateRegistryRecord on the target registry. AWS rejects unauthorized publication. Triggers on phrases like "publish this skill", "register this in our skill registry", "把这个 skill 发布出去", "推到 skills registry".

FileKindSize
SKILL.mdskill md4.8 KB
resources/publish-skill-runbook.mdmarkdown2.7 KB
scripts/publish.pyscript5.1 KB

resources/publish-skill-runbook.md

View on GitHub
# Publish-skill runbook — GA

The interpreter must have the shared blueprint package installed with publishing extras:
`python -m pip install -e '.[publish]'` from the cloned repository.
Copying this skill's directory does not copy that library.

## Preconditions

- Python 3.11+; a directory with `pyproject.toml` and `src/<normalized_package>/skill_files/SKILL.md`.
- A GA Registry and CodeArtifact repository, plus explicit publishing authorization.
- The publisher role must not have Registry approval permissions.

## 1. Local preflight

```bash
python <path-to-publish.py> --package-dir <skill-project> --dry-run
```

This performs no AWS calls and no build. It is not an IAM permission audit.
Defaults are `skills-demo`, `skills-prod`, `skills-demo-registry`, `us-east-1`;
override them with flags or `~/.skillpublish/config.toml`.
`AGENT_REGISTRY_ARN`/`--registry-id` avoids listing registries by name.

## 2. Publish

After the user approves publication:

```bash
python <path-to-publish.py> --package-dir <skill-project>
```

Builds one wheel, derives its Markdown/digests, uploads it, verifies the CodeArtifact hash,
and creates a `SKILL` record in `DRAFT`. No package installation hook executes.
Twine receives a short-lived token through environment variables, not a persistent config file.

For an existing wheel use `--wheel path/to/release.whl`.
If upload already succeeded, retry with `--wheel ... --skip-upload` so the same bytes are reused.
The remote digest is still checked. Bump the version for different content.

## 3. Inspect and submit

```bash
aws agent-registry-control get-registry-record \
  --registry-id REGISTRY_ARN --record-id RECORD_ID --region us-east-1
aws agent-registry-control submit-registry-record-for-approval \
  --registry-id REGISTRY_ARN --record-id RECORD_ID --region us-east-1
```

Check `recordType=SKILL`, `recordVersion`, `agentSkillsDefinition.data`,
the `com.example.integrity`/`com.example.codeartifact` conventions, and
`agentSkillsDefinition.additionalData.skillMd.data`.
Alternatively, `--auto-submit` submits immediately after creation/reuse of a draft.

## 4. Separate curator

Approval is outside this publishing skill's scope.
The curator reviews content and provenance and uses `scripts/03_approve_skill.py --reason ...`.
Allow discovery indexing to catch up after approval.

## 5. Verified consumer

Use `scripts/04_consume_skill.py` for the approved exact version.
Do not translate record metadata into arbitrary shell commands, run the legacy postinstall command,
or recommend unverified `pip install` as the trusted consumption flow.

For permissions, limitations and live acceptance steps, see `docs/09-publishing-iam.md`,
`docs/12-record-artifact-integrity.md`, and `docs/03-demo-walkthrough.md` in the repository.
writing-skillsobra/superpowersUse when creating new skills, editing existing skills, or verifying skills work before deploymentStars: 322kUpdated Claude CodeCodexVerifiedInstall
xlsxanthropics/skillsUse this skill any time a spreadsheet file is the primary input or output. This means any task where the user wants to: open, read, edit, or fix an existing .xlsx, .xlsm, .xltx, .csv, or .tsv file (e.g., adding columns, computing formulas, formatting, charting, cleaning messy data); create a new spreadsheet from scratch or from other data sources; or convert between tabular file formats. Trigger especially when the user references a spreadsheet file by name or path — even casually (like "the xlsx in my downloads") — and wants something done to it or produced from it. Also trigger for cleaning or restructuring messy tabular data files (malformed rows, misplaced headers, junk data) into proper spreadsheets. The deliverable must be a spreadsheet file. Do NOT trigger when the primary deliverable is a Word document, HTML report, standalone Python script, database pipeline, or Google Sheets API integration, even if tabular data is involved.Stars: 198kUpdated Claude CodeCodexVerifiedInstall
shipping-and-launchaddyosmani/agent-skillsPrepares production launches. Use when preparing to deploy to production, or when asking what needs to be in place before shipping. Use when you need a pre-launch checklist, when setting up monitoring, when planning a staged rollout, or when you need a rollback strategy.Stars: 119kUpdated Claude CodeCodexVerifiedInstall
childhoodthedotmack/claude-memBabysit a shipped release after merge, publication, or deployment. Verify published packages, production behavior, installed plugins, and late CI or user reports until the observation window is clean; use babysit for PR readiness before release.Stars: 118kUpdated Claude CodeCodexVerifiedInstall