deploy-setup
Skill2 starsChanged 2 months agoMITView on GitHub
Wire up a freshly scaffolded Starter Series project for its first release — detect the template, register required GitHub secrets, set up OIDC trusted publishing where applicable, and trigger the first CD run. Pairs with the `create` skill (you scaffold first, then run this).
- Reads credentials
- Commits and pushes
1 other project uses a version of this.
Adopted by
Most recently taken up by
- rootkit-lab/skinktheir own version, first seen Aug 12, 2026
Other versions1 project uses a version of this.
Word for word · 0
None found.
Edited · 0
None found.
Wrote their own version · 1
- rootkit-lab/skink★ 0same name
How matches are found. Word for word: the exact same text. Edited: shares passages with this file (“99% same” means 99% of its five-word phrases also appear here; 10% or more counts). Their own version: similar meaning, or, weakest, the same name. We can’t tell which project had it first. Projects are counted among the 200 most-starred candidates, and mentions are links to the repository we found, up to 50 per site, so both counts are a floor.
In this repositoryOne other file it gives its agents.
- createSkill
The file
We split this skill into the rules it gives an agent
9 rules. About 0 are common, 3 are distinctive, and 6 are very specific to this project.
≈0
Common · estimate
Many files like this have a rule of this kind.
≈3
Distinctive · estimate
This author's own choices. Worth a look.
6
Very specific
Tied to their project's names, files or tools. Probably skip.
- Common
- Distinctive
- Very specific
- Distinctive: line 18
Read these files to identify which starter the project came from:
- Very specific: line 38
Read .github/workflows/cd.yml (or cd-android.yml + cd-ios.yml for react-native, or cd-firefox.yml for browser-extension).
- Distinctive: line 45
If the user wants verbatim guidance, use this table:
- Very specific: line 63
Run gh secret list -R <owner>/<repo> to see what's already set.
- Distinctive: line 112
After the run finishes:
- Very specific: line 125
If the user asks for a dry run, do everything except gh secret set and gh workflow run.
- Very specific: line 136
Existing failed run: if there's a failed run on cd.yml, surface the failure with gh run view <id> --log-failed | head -50 so we don't …
- Very specific: line 137
User aborts mid-flow: do not auto-retry. Save progress in a one-line .deploy-setup.checkpoint file at repo root listing which secrets/OIDC …
- Very specific: line 141
Never: put a token value in a chat message you produce.
≈ “Common” is an estimate: at least 1 in 5 of the 350,495 skill files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.
Files
You are walking a developer through the first deploy of a Starter Series project. The project has been scaffolded (via create-starter skill / CLI / MCP) and is on disk; this skill takes it from "scaffolded" to "first release published."
Prerequisites
- The project is already scaffolded and on disk. Verify by checking for
.github/workflows/cd*.ymlandpackage.json/pyproject.toml/Dockerfile. ghCLI is installed and authenticated (gh auth status). If not, instruct the user.- A GitHub repository exists for the project. If not, offer
gh repo create.
Workflow
Step 1: Detect the template
Read these files to identify which starter the project came from:
| Signal | Template |
package.json#mcpName starts with io.github. and package.json#main exists, no electron dep | mcp-server (Node MCP) |
pyproject.toml references mcp package | mcp-server-python |
package.json#dependencies.discord.js | discord-bot |
package.json#dependencies.grammy | telegram-bot |
manifest.json#manifest_version === 3 | browser-extension |
package.json#engines.vscode | vscode-extension |
package.json#dependencies.electron (or devDependencies.electron) | electron-app |
package.json#dependencies.expo | react-native |
wrangler.toml exists | cloudflare-pages |
Dockerfile at root + no package.json | docker-deploy |
package.json#publishConfig.access === "public" and no other signal | npm-package |
Report the detected template name to the user.
Step 2: Inspect the CD workflow
Read .github/workflows/cd.yml (or cd-android.yml + cd-ios.yml for react-native, or cd-firefox.yml for browser-extension). Extract:
- The list of
secrets.Xreferenced inwith:/env:/if:blocks. These are the secrets that must be set. - Whether the workflow uses OIDC trusted publishing (look for
id-token: writeinpermissions:andnpm publish --provenance/pypa/gh-action-pypi-publishin steps). If yes, the publish step needs no token secret — only registry-side configuration (web UI). If no, a token secret is required.
Step 3: Required secrets matrix (per template)
If the user wants verbatim guidance, use this table:
| Template | Required GitHub secrets | Registry-side setup |
mcp-server | (none — OIDC via npm) | npm: enable trusted publisher for the package |
npm-package | (none — OIDC via npm) | npm: enable trusted publisher |
mcp-server-python | (none — OIDC via PyPI) | PyPI: register trusted publisher under https://pypi.org/manage/account/publishing/ |
vscode-extension | VSCE_PAT (required), OVSX_PAT (optional) | Azure DevOps: create PAT with Marketplace > Manage scope |
browser-extension | CHROME_REFRESH_TOKEN, CHROME_CLIENT_ID, CHROME_CLIENT_SECRET, CHROME_EXTENSION_ID, AMO_JWT_ISSUER, AMO_JWT_SECRET | CWS: OAuth client. AMO: API key |
electron-app | GH_TOKEN (or use GITHUB_TOKEN); platform code-signing certs (optional) | None for first run; signing certs for release |
react-native | EXPO_TOKEN, APPLE_ID, APPLE_APP_SPECIFIC_PASSWORD, APPLE_TEAM_ID, ASC_API_KEY_*, ANDROID_KEYSTORE_BASE64, etc. | Expo: account; App Store Connect API key; Play Console service account |
cloudflare-pages | CLOUDFLARE_API_TOKEN, CLOUDFLARE_ACCOUNT_ID | Cloudflare: API token with Pages:Edit |
discord-bot | RAILWAY_TOKEN + RAILWAY_SERVICE_ID (Railway path) or FLY_API_TOKEN (Fly path) | Railway: project + service; Fly: app create |
telegram-bot | Same as discord-bot | Same |
docker-deploy | VPS_HOST, VPS_USER, VPS_SSH_KEY, APP_PORT (optional) | VPS reachable; ~/.env.app exists on host |
Step 4: Diff against existing secrets
Run gh secret list -R <owner>/<repo> to see what's already set. Subtract from the required list. For each missing secret, ask the user:
"I need a value for
VSCE_PAT. This is a Personal Access Token withMarketplace > Managescope from https://dev.azure.com/_usersSettings/tokens. Paste the token (it will be masked):"
Then run gh secret set VSCE_PAT --body "$value" -R <owner>/<repo>.
Never echo the value to the conversation, never write it to disk, never include it in git commit messages.
Step 5: OIDC trusted publishing setup (where applicable)
For mcp-server, npm-package, create-starter, mcp-server-python:
The CD workflow requires registry-side trusted publisher registration before the first publish. Provide the user with:
- npm: link to https://www.npmjs.com/settings/
<their-org-or-user>/packages → select package → Trusted Publishers → "Add GitHub Actions" with the repo + workflow filename. Tell them: "Your npm account must have 2FA on writes enabled for trusted publishing to be available." - PyPI: link to https://pypi.org/manage/account/publishing/ → "Add a new pending publisher" with the GitHub org, repo name, workflow filename, and (optionally) environment name.
These steps cannot be automated via gh — they're registry-side web UI. Provide the URL and exact field values.
Step 6: Trigger the first deploy
Once secrets are in place and OIDC is registered:
# Bump version first (some templates require this; check cd.yml's version-guard step)
Existing-repository setup may identify former templates. For new template downloads, use `list_templates`; retired templates are unavailable for scaffolding and `add_component`.
npm run version:patch # or version:minor, or version:major
# Commit + push the version bump
git commit -am "chore: bump version for first release"
git push
# Trigger the CD workflow
gh workflow run cd.yml -R <owner>/<repo>
# Watch the run
gh run watch -R <owner>/<repo>For tag-triggered repos (docker-deploy uses push: tags: v*), instead:
git tag v0.1.0
git push origin v0.1.0Step 7: Verify the release
After the run finishes:
gh release view -R <owner>/<repo>For npm: npm view <package-name> shows the published version.
For PyPI: pip index versions <package-name> (or visit pypi.org/p/<name>).
For Marketplace: visit https://marketplace.visualstudio.com/items?itemName=<publisher>.<name>.
For GHCR: docker pull ghcr.io/<owner>/<repo>:<version>.
Dry-run mode
If the user asks for a dry run, do everything except gh secret set and gh workflow run. Print the would-be commands and the secret names you'd request. Useful for review before committing to a deploy.
Output format
Walk the user through one numbered step at a time. After each interactive prompt, summarize what got set / what's pending. End with a "Next: trigger cd.yml" once secrets and OIDC are in place.
Edge cases
- Repo not yet created: offer
gh repo create <name> --public --source=. --pushand re-detect after. - Multiple cd workflows (react-native, browser-extension): detect both, run their secret matrices in sequence.
- GH_TOKEN already a default: GitHub Actions provides
secrets.GITHUB_TOKENautomatically with read+write to the same repo. Don't ask the user to set it. - Existing failed run: if there's a failed run on
cd.yml, surface the failure withgh run view <id> --log-failed | head -50so we don't blindly retrigger. - User aborts mid-flow: do not auto-retry. Save progress in a one-line
.deploy-setup.checkpointfile at repo root listing which secrets/OIDC steps are done, and resume from there next invocation. (Optional — only if user asks.)
Safety rails
- Never put a token value in a chat message you produce. Always pipe to
gh secret set --body "$VALUE". - Never modify CD workflows during this skill — it's setup, not editing. If a workflow looks wrong, surface it but ask the user before touching.
- Never publish without the user explicitly confirming "yes, run cd.yml now."
- For first npm/PyPI publishes, recommend the user manually trigger once with workflow_dispatch (auditable) before any tag-based automation.
Does it work?
Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.
No reviews yet
People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.
- It worked
- Partly
- It didn't
Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.
Your trail
Files you open will collect here, so you can find them again.
Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.
Now do this for your own.
Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.

