land-and-deploy
by garrytanland-and-deploy is a workflow skill for merging a ready PR, waiting for CI and deployment, and verifying production health with canary checks. Use this land-and-deploy skill when you need a reliable handoff from merge to ship, especially for teams that already use /ship or similar release steps.
This skill scores 68/100, which means it is listable but best presented with caution. It gives directory users a real, triggerable land-and-deploy workflow with explicit merge/deploy/verify intent, but the repository still has some discoverability and trust gaps that make it less self-explanatory than a stronger install candidate.
- Explicit trigger coverage for deployment handoff cases like "merge and deploy," "land the pr," and "ship to production".
- Substantial operational content: the body is large and includes many workflow and constraint signals, suggesting a non-trivial executable process rather than a placeholder.
- Includes concrete execution scaffolding in the SKILL.md preamble, including branch/session checks and repo-mode setup, which should help an agent act with less guesswork.
- The frontmatter description is very short and the repository has no supporting scripts, references, resources, or README, so users have limited install-time context.
- Placeholder markers such as todo/wip/placeholder appear in the skill content, which raises some confidence risk about completeness or polish.
Overview of land-and-deploy skill
What land-and-deploy does
land-and-deploy is a workflow skill for the handoff from a merged PR to a verified production deployment. It is built for the land-and-deploy step of a release process: merge the change, wait for CI and deployment, then confirm the app is healthy with canary checks. If you need a land-and-deploy skill that reduces guesswork around the final release step, this is aimed at that job.
Who should install it
Install land-and-deploy if you routinely ask an agent to “merge it,” “land the PR,” “ship to production,” or “merge and verify.” It is most useful for operators who already have a /ship or PR-creation step and want a follow-through skill for the deploy-and-check phase. It is less useful if your team does not allow agents to merge, deploy, or validate production at all.
What makes it different
The main value is not generic release advice; it is a constrained deployment workflow with explicit attention to safe operations, CI/deploy waiting, and post-deploy verification. The land-and-deploy skill also includes routing and preamble behavior, which helps the agent recognize when the workflow should take over instead of improvising a broad answer.
How to Use land-and-deploy skill
Install context and trigger phrases
Use the land-and-deploy install flow in your skill manager, then invoke it when the task is specifically about landing a change and confirming production health. The repository description and triggers point to phrases like merge and deploy, land the pr, ship to production, and merge and verify. In practice, a strong land-and-deploy usage prompt should say what branch or PR is ready, what environment is expected, and what “done” means.
Best first files to read
Start with SKILL.md, then inspect SKILL.md.tmpl to see the templated source that generates the behavior. Because this repo has no supporting rules/, references/, resources/, or scripts/ folders, the main implementation lives in the skill file itself. For land-and-deploy guide reading, focus on the preamble, safe-operation constraints, plan-mode behavior, and routing sections before anything else.
How to prompt for a good run
Give the agent the minimum release facts it needs to act safely: repository name, PR link or branch, deploy target, any required approvals, and the health signal you want checked after deployment. A better prompt is specific, for example: “Use land-and-deploy to merge PR #184, wait for CI and production deployment, then confirm the canary checks pass in staging-to-prod rollout.” Avoid vague asks like “ship it” unless the repo context already makes the target obvious.
Workflow tips that matter
The skill is designed around taking over after PR creation, so do not ask it to invent the change itself. If the deployment path depends on repo-specific commands, branch naming, or a required verification URL, include those up front. For land-and-deploy for Deployment, the quality jump usually comes from clarifying the acceptance check: smoke test, canary, status page, or another production-health signal.
land-and-deploy skill FAQ
Is this just a generic deploy prompt?
No. land-and-deploy is a focused operational skill for the merge-to-production phase, with explicit support for waiting on CI, waiting on deploy, and verifying health. A normal prompt can describe that sequence, but the skill is meant to make the agent follow it consistently.
When should I not use it?
Do not use land-and-deploy if you need design decisions, code implementation, or a release plan that is not yet ready to merge. It is also a poor fit if your org prohibits automated merging or production actions, or if you need a manual-only checklist with no agent execution.
Is it beginner-friendly?
Yes, as long as the user can state the target repo, PR, and deployment expectation. The skill reduces release ambiguity, but it still depends on the operator knowing what should be landed and how success is measured. Beginners usually get the best results when they provide the PR and ask the agent to follow the land-and-deploy workflow exactly.
What should I compare it with?
Compare land-and-deploy with a plain “merge and deploy” prompt. The skill is better when you want repeatable behavior, consistent safe-operation boundaries, and a clearer path from merge to verification. If you only need one-off instructions, a prompt may be enough; if this is a recurring release step, the skill is the stronger install decision.
How to Improve land-and-deploy skill
Give stronger release inputs
The best way to improve land-and-deploy output is to specify the exact PR, branch, environment, and verification method. Instead of “deploy this,” say “land PR #57 to main, wait for the production pipeline, then confirm /health returns 200 and the release canary is green.” That gives the skill a concrete finish line.
Reduce failure modes early
Common failure modes are missing deploy context, unclear approval boundaries, and weak post-deploy checks. If your workflow has protected branches, required reviews, feature flags, or manual rollout steps, name them before the skill acts. This avoids the agent guessing around policy constraints or stopping too early after merge.
Iterate after the first run
If the first land-and-deploy pass is too cautious, too broad, or checks the wrong signal, refine the prompt with the missing operational detail rather than restating the goal. Add repo-specific commands, links to the deploy dashboard, or the exact canary criterion you want verified. That is the fastest way to make land-and-deploy behave like your real release process instead of a generic deployment script.