deploy-workflow
SkillCloud & infraRelease deployment process for ccswarm. Version update, quality gates, build, tag, publish.
Available today. Use it from your connected AI after setup.
No other account needed.
Connect ahel once, and every AI you use reads what you have installed.
Then ask your AI: use the deploy-workflow skill
What this skill tells your AI
The instructions your AI receives, as published by nwiizo/ccswarm in .agents/skills/deploy-workflow/SKILL.md and read by ahel’s review.
Deploy a new ccswarm release using .github/workflows/release.yml.
The workspace has one published crate, crates/ccswarm. Follow the scope in
docs/ROADMAP.md and the version-specific notes in docs/releases/.
Pre-Deployment
# Update crates/ccswarm/Cargo.toml and its Cargo.lock package version.
# Update CHANGELOG.md and docs/releases/v<VERSION>.md.
# Quality gates
cargo fmt --all
cargo clippy --workspace --locked -- -D warnings
cargo test --workspace --locked
cargo audit
cargo doc --no-deps --workspace --locked
cargo build --release --workspace --locked
cargo publish -p ccswarm --dry-run --locked
actionlint .github/workflows/release.yml
# Browser example (installs its locked dependencies and Chromium if needed)
CCSWARM_BIN=target/release/ccswarm examples/e2e-playwright/run.sh
Build & Release
When release publication is authorized, verify the branch, remote, diff, and unused tag. Commit and push the reviewed release preparation. Run the Release workflow manually on that exact commit and verify all four native archive builds, smoke tests, and quality checks. Manual runs do not publish.
Then create and push the annotated version tag on the verified commit. The tag workflow rebuilds and validates, creates a complete draft, verifies its downloaded assets, attests the archives, publishes the stable crate, and publishes the GitHub release. Prerelease tags do not publish a crate.
Do not run a second manual publish command alongside the workflow. Verify the completed workflow, all four archives and checksums, provenance, registry version, and an installed binary before reporting success.
Failed Publication And Recovery
Inspect the failed job and current GitHub/crates.io state before retrying. The two services cannot publish atomically. A workflow retry verifies an existing crate's checksum against the candidate before skipping its upload. Investigate a mismatch rather than bypassing that check. Do not replace an already published tag or delete a published release as routine recovery. Use a follow-up version for code fixes; yanking a broken crate requires explicit authorization.
Signals
- GitHub stars
- 151
- Forks
- 15
- Last commit
- Sep 2026
Advanced
- Catalog kind
- skill
- Gateway key
deploy-workflow- Source
- github.com/nwiizo/ccswarm