ry-deploy

ry-deploy

Purpose

Synchronize local repository, GitHub, and server, then deploy safely with evidence. Automation is expected, but risky operations require code/log evidence and user questions with options.

Workflow

1. Read deploy contract from AGENTS.md, then .claude/CLAUDE.md, then .serena/deploy/*.md. See ${CLAUDE_PLUGIN_ROOT}/references/deploy-contract.md for the required field set.
2. Run bash ${CLAUDE_PLUGIN_ROOT}/scripts/deploy_readiness.sh <server> when available.
3. Verify local git state, open PR, checks, Serena memories, docs, and GitHub sync.
4. Inspect server baseline: git status, current commit, disk, logs before restart, process manager.
5. Sync code to server.
6. Run migrations only when readiness is clear.
7. Restart/build services.
8. Verify logs, tests, health checks, and critical behavior.
9. If anything fails, perform root cause analysis using server logs, code, and internet research. Fix-forward and redeploy. Ask the user with options for risky or ambiguous decisions.
10. DB rollback only when explicit rollback command and backup/restore point are verified.
11. Finish with flow-post-task-sync.

No Fake Success

If auth, missing credentials, server access, or unavailable health checks prevent validation, state the limitation and what evidence was still collected.

Anti-patterns

• Restart services без чтения logs до restart (нет baseline для compare).
• Migrations без verified backup/rollback contract.
• DB rollback без explicit rollback command + verified restore point.
• Fake success report когда auth blocked validation.
• Skip flow-post-task-sync after deploy.