ci
/ci
Podłącza ciągłe wdrażanie repozytorium do Gagarin Cloud po każdym pushu, z ograniczonym kluczem wdrożeniowym i plikiem workflow.
- Autor
- gagarin-cloud ↗
- Repozytorium
- plugin ↗
- Licencja
- MIT
- Ostatnia zmiana repozytorium
- 13.11.2026
Ocena i bezpieczeństwo
- Pliki w paczce
- 1
- Skrypty
- 0
- Komentarze HTML
- 0
- Adresy URL w SKILL.md
- 0
Brak sygnałów. Repozytorium zmienione 13.11.2026, bez odwołań do wycofanych modeli i API.
MIT, otwarta: można używać i zmieniać, z zachowaniem informacji o autorze.
Co robi i kiedy
Podłącza ciągłe wdrażanie repozytorium do Gagarin Cloud po każdym pushu, z ograniczonym kluczem wdrożeniowym i plikiem workflow.
Oryginalny opis (EN)
Wire a repository's CI pipeline to deploy to Gagarin Cloud on push — mint a scoped deploy credential, store it as a secret, and write the workflow. Use when the user asks to set up continuous deployment, automate deploys, or deploy from GitHub Actions or another CI system to Gagarin.
Jak działa
- The order
- Rules that are not negotiable
- What the credential can and cannot do
- Finishing
Podgląd SKILL.md
---
name: ci
description: Wire a repository's CI pipeline to deploy to Gagarin Cloud on push — mint a scoped deploy credential, store it as a secret, and write the workflow. Use when the user asks to set up continuous deployment, automate deploys, or deploy from GitHub Actions or another CI system to Gagarin.
---
You wire this repository's CI to deploy to Gagarin Cloud on push.
Read the `gagarin` skill of this plugin before anything else: its **"Setting up
CI"** section is the recipe. What follows is the order of work and the rules
you must not break. If your host offers subagents, run the rest of this in one:
it is a long procedure, and the main conversation only needs its report.
Everything here runs on the machine that holds the repository. If you cannot
run commands where you are (ChatGPT on the web, for instance), hand the user
the recipe instead — the `gg creds create` line and the workflow — and say
which secret to set where. Do not try to do it over the MCP tools: no tool
there hands out a secret, by design.
## The order
1. **`gg whoami`** — you need a working credential to mint one from. If there is
none, stop and tell the user to run this plugin's `setup` skill
(`/gagarin:setup` in Claude Code).
2. **Find out what CI they use.** Look for `.github/workflows/`, `.gitlab-ci.yml`,
`.buildkite/`, `cloudbuild.yaml`. Ask if it is ambiguous. Do not assume GitHub.
`git remote get-url origin` says where the repository lives.
3. **Check the forge CLI.** On GitHub that is `gh auth status`; on GitLab,
`glab auth status`. If it is missing or not signed in, say so before minting
anything: ask the user to install it and run `gh auth login` / `glab auth
login` themselves, or to set the secret by hand (see below). Do not sign in
for them.
4. **Confirm the project and service names** and which branch should deploy.
Read `.gagarin.json` at the repository root first — the note of which
project this is, if a deploy left one — then `gg status` or `gg projects`
rather than guessing. If the two disagree, the API is right. Do not write
the note here if it is absent: that is the deploy's job, and there may be
no project yet.
5. **Mint a credential straight into the secret store**, named for where it
will live (`"github actions: <owner>/<repo>"`) — the one-line pipe in the
rule below, never `gg creds create` on its own, which prints it.
6. **Write the workflow.** Install gg, `gg registry login`, build and push
tagged with the commit, deploy, then `gg status`.
7. **Tell them what to check** on the first run — and, once they have pushed,
watch it with the forge CLI (the skill's "Watch the first run").
## Rules that are not negotiable
- **Never run `gg login` in CI, and never copy this machine's credential into
it.** A pipeline gets one of its own.
- **The minted secret must never reach the transcript, a file, or your output.**
`gg creds create` prints it to stdout alone on its line, everything else to
stderr, precisely so it can be piped without touching disk:
```sh
gh secret set GAGARIN_TOKEN --body "$(gg creds create --name "github actions: acme/web" 2>/dev/null)"
```
On GitLab, the same shape:
```sh
glab variable set GAGARIN_TOKEN --masked --value "$(gg creds create --name "gitlab ci: acme/web" 2>/dev/null)"
```
Run it as one command. Do not capture it into a shell variable you later echo,
do not print it "to confirm", and do not write it into a file — not even a
temporary one. It is shown once; Gagarin keeps only a hash. If the user needs
it again they revoke and mint another.
- **No signed-in forge CLI means you do not mint it.** There is no route from
here to the secret store that skips your output. Hand the user the
`gg creds create --name "…"` command to run in their own terminal and say
where in the CI settings to paste it.
- **`gg registry login` is not optional** before any `gg build --push`,
`gg push` or `gg registry copy`. Docker has no idea what a Gagarin credential
is until it runs, and the failure without it is a `401` out of `docker push`.
- **Tag with the commit, explicitly.** `${{ github.sha }}` is what makes a
deploy reproducible and a rollback explicable. `gg build` with no tag uses
the short hash only when the checkout is clean and falls back to the clock
otherwise, and a pipeline that writes a file before building is not clean —
so never leave the tag to it in CI.
- **`gg status` at the end is a report, not a gate.** Do not write a polling
loop in CI. If they want a gate, gate on their own health check against the
service's address.
## What the credential can and cannot do
Say this to the user explicitly, because it is the reason this is safe: what you
minted **can deploy but cannot destroy** — no flag turns that on — it **expires**
(90 days by default, 365 maximum, never "never"), and it **cannot mint another**.
Tell them the expiry date and that rotating means minting the new one, updating
the secret, then `gg creds revoke <id>`.
## Finishing
Report which secret you set, where, what the workflow does on which branch, and
when the credential expires. Do not trigger a deploy to test it unless the user
asks — that is their push to make. If they push while you are here and the
forge CLI is signed in, watch the run to the end and report how it finished,
with the failing step's log if it failed, then `gg status`.
Wersje
| Repozytorium | Wynik | Licencja |
|---|---|---|
| Ta strona · gagarin-cloud/plugin | 100 | MIT |
| A7-studio/jobmigo | 100 | MIT |
| andriiklymiuk/corgi | 100 | MIT |
| kamoca-labs/docker | 100 | MIT |
| pawangai/pawang-agents | 100 | MIT |
| agentic-dev3o/devx-plugins | 100 | MIT |
| chachamaru127/claude-code-harness | 100 | MIT |
| onsi/ginkgo | 100 | MIT |
| vale-cli/agent-tools | 100 | MIT |
Migawka bazy z 13.11.2026 · wersja paczki 6b420b507fc9 (skrót SHA-256 całej paczki)
Pobrania w ostatnich 30 dniach: 0 · ranking
Zgłoś problem z tym skillem: szkodliwa treść, błędny opis albo prośba autora o usunięcie.
Podobne skille
- 90setup
Przygotowuje ten komputer do wdrożeń w Gagarin Cloud: instaluje CLI gg, loguje się i sprawdza narzędzia Docker, Git oraz gh lub glab. - 91gagarin
Wdraża i obsługuje aplikacje na platformie Gagarin przez CLI gg: usługi, bazy danych, połączenia, zadania i logi. - 95memstack-deployment-ci-cd-pipeline
Konfiguruje automatyczne pipeline'y budowania, testów i wdrożeń w GitHub Actions. - 100fastapicloud-deploy
Przygotowuje i wdraża aplikacje FastAPI do FastAPI Cloud: logowanie, CI/CD w GitHub Actions, tokeny wdrożeniowe, walidacja wejść, .fastapicloudignore i polecenia fastapi deploy. - 100netlify-deploy
Tworzy i zarządza wdrożeniami Netlify z kodu: ciągłe wdrażanie z Git, podglądy, hooki i naprawa nieudanych deployów. - 100github-actions
Obsługuje GitHub Actions CI/CD: workflowy, zadania i wdrożenia do automatyzacji budowania, testów i wdrożeń.