Deploy

Skill lantisprime/claude-sdlc/skills/deploy

A governance layer for AI-assisted software delivery. Plan, scope, approve, build, test, release — with human sign-off at every gate.

Install

npx -y skills add lantisprime/claude-sdlc --skill deploy

Assembled from the repository path, not quoted from the project. Check it against their README if it does not work.

One thing to look at

3 stars3 stars. Stars are a popularity signal and not a quality one, but at this level it is likely that nobody has read this closely except its author, and you would be relying on your own review.

What its author says it does

Copied from the file, not written here

Use this skill during Phase 6 to deploy code that has passed Phase 5 Test. Updates tickets with deployment details when a ticketing system is configured; otherwise generates deployment artifacts as markdown or JSON under .claude/sdlc/deployments/. Never auto-executes a deployment — always proposes and waits for explicit human confirmation. Trigger after test is signed off, or when the user says "deploy", "release", or "ship".

SKILL.md

3.2 KB, as published. Nobody here has run it

Deploy (Phase 6)

Release code that has passed Test — under explicit human approval.

Prerequisite

.claude/sdlc/gates/test-<task-slug>.md must exist and be signed. Without it, this skill refuses to run.

Step 1 — Propose the deployment

Produce a deployment proposal, not a deployment. Include:

Target environment (dev / staging / production)
Commit SHA / tag / image reference
Changed files and services
DB migrations? (yes/no, reversible?)
Feature flags? (default state)
Rollback plan
Blast radius (users/services affected)
Deployment window / maintenance window if applicable

Step 2 — Human approval gate

This is non-negotiable. Deployments never auto-execute. Ask the human to confirm the proposal. Record the approval at .claude/sdlc/gates/deploy-<task-slug>.md with timestamp and approver.

Step 3 — Execute (under human supervision)

Execute the deployment via the configured pipeline/runbook. Stream output. Halt on the first error and ask for direction — do not retry automatically.

Step 4 — Record

Update the ticket(s) with:

Commit SHA, environment, timestamp, approver
Link to pipeline run / deployment log
Post-deploy verification results (smoke tests)

Detection:

GitHub/GitLab Issues via .claude/sdlc/env.json → comment on the issue
Jira/Linear via MCP → transition to "Deployed" (or equivalent) with a comment
None configured → write .claude/sdlc/deployments/<YYYY-MM-DD>-<task-slug>.md (or .json if configured)

Step 5 — Post-deploy verification

Run the smoke-test subset of the test suite against the deployed environment. If any fail:

Execute the rollback from Step 1
Log a defect (see test skill)
Do not mark the deployment complete

What this skill must NEVER do

Auto-execute a deployment.
Skip the rollback plan.
Close a ticket before post-deploy verification passes.
Deploy code that does not have a signed test gate.

References

templates/deployment.md
templates/gate.md
docs/SDLC.md Deploy

Next step hint

After the deployment record is written, pipe the next_suggestions conditions to skills/_shared/next-hint.sh and print any output:

printf '%s\n' \
  'deploy_gate_signed|run /support to wire observability for the deployment' \
  'phase_active_gate_unsigned|open gates/deploy-<task-slug>.md and sign it manually, then run /support' \
  | bash skills/_shared/next-hint.sh

Print any output verbatim. If the script outputs nothing, add nothing.