agentsclimarketplace

Run deploy

Skill mataeil/OODA-loop/skills/run-deploy

Trigger deployment via GitHub Actions workflow_dispatch. Execute phase skill — validates pre-deploy conditions, fires the configured workflow, monitors completion, and verifies post-deploy health.From its SKILL.md

Install
npx -y skills add mataeil/OODA-loop --skill run-deploy

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

2 things to look at

  • 6 stars6 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.
  • runs commandsInstructs the agent to run 7 commands, including `gh --version` and 6 more.

SKILL.md

8.1 KB, ~1.8k tokens by cl100k_base, as published. Nobody here has run it

run-deploy: GitHub Actions Deployment Trigger

The "hand" of the harness. Validates deployment readiness, fires the configured GitHub Actions workflow via workflow_dispatch, monitors completion, and verifies the service is healthy afterward.

  • NEVER auto-invoked without passing the full pre-deploy checklist
  • Requires explicit conditions — does not deploy blindly
  • Writes results to agent/state/deploy.json

Safety Rules

  1. HALT file — Mandatory first check. If present, print reason and stop.
  2. Pre-deploy checklist — All six conditions must pass before triggering.
  3. gh required — Cannot deploy without GitHub CLI. If unavailable, error and exit.
  4. No blind deploys — config.deploy_workflow must be explicitly configured.

Step 0: Safety

0-A: HALT Check

if file exists at config.safety.halt_file:
  Print "[HALT] run-deploy stopped. Reason: {file_content}"
  EXIT immediately.

0-B: Config Validation

if config.deploy_workflow is missing, null, or empty:
  Print "No deploy workflow configured. Skipping."
  EXIT cleanly (not an error).

Step 1: Pre-Deploy Checklist

All six conditions must pass. On first failure, print the reason and skip deployment.

#CheckSourcePass Condition
1No critical health alertsagent/state/service_health.jsonalerts has no item with severity == "critical"
2Tests passingagent/state/test_coverage.jsonstatus == "passing"
3No HALT fileconfig.safety.halt_fileFile does not exist (re-confirmed)
4Complexity levelconfig.jsonprogressive_complexity.current_level >= 3 OR invoked directly by user
5Clean working treegit status --porcelainOutput is empty (no uncommitted changes)
6Workflow file exists.github/workflows/{deploy_workflow}File exists in the repository (prevents cryptic GitHub API errors)

On any failure:

Pre-deploy checklist FAILED: {reason}
Deployment skipped.

State files missing (health, tests) count as unknown — treat as failure for that check. Print which specific check failed so the user can act on it.


Step 2: Trigger Deployment

Verify gh is available and authenticated:

gh --version
gh auth status

If not found: print ERROR: gh (GitHub CLI) is required for deployment. Install from https://cli.github.com/ and exit non-zero. If not authenticated: print ERROR: gh is not authenticated. Run: gh auth login and exit non-zero.

Determine the current branch:

git rev-parse --abbrev-ref HEAD

Trigger the workflow. If config.deploy_workflow_inputs is set (object of key-value pairs), pass each entry as -f key=value:

# Without inputs:
gh workflow run {config.deploy_workflow} --ref {branch}

# With inputs (e.g. {"environment":"staging","tag":"v1.2.3"}):
gh workflow run {config.deploy_workflow} --ref {branch} -f environment=staging -f tag=v1.2.3

Wait up to 15 seconds for the run to register, then retrieve the run ID:

gh run list --workflow={config.deploy_workflow} --limit=1 --json databaseId,url --jq '.[0]'

Print:

Deployment triggered: {config.deploy_workflow} on {branch}
Workflow run: {run_url}

Step 3: Monitor

Watch the run until completion. Timeout is read from config.deploy_monitor_timeout_seconds (default 600 — 10 minutes):

gh run watch {run_id} --exit-status
  • Exit 0 → status: "success"
  • Non-zero exit → status: "failed"
  • Timeout (> configured seconds): kill watch, record status: "unknown", print: Deployment monitor timed out after {timeout}s. Check {run_url} manually.

Re-check the HALT file after the wait completes. If it appeared during the long monitor window, record status: "halted" and skip post-deploy steps.


Step 4: Post-Deploy Health Check

Skip this step if config.health_endpoints is missing or empty — note the skip.

If endpoints are configured:

  1. Wait config.deploy_health_wait_seconds (default 30) for the service to stabilize.
  2. Run health checks using the same logic as scan-health (Step 2 of that skill).
  3. If any endpoint fails, retry once after another deploy_health_wait_seconds wait (services with rolling deploys may need extra time).
  4. Evaluate final results:
    • All endpoints 200 → health_check: "passed" — print Deployment verified healthy.
    • Any endpoint non-200 or critical alert after retry → health_check: "failed" — print: ALERT: Post-deploy health check failed after retry. Consider rollback.

Step 5: State Update + Report

Write to agent/state/deploy.json:

{
  "schema_version": "1.1.0",
  "last_deploy": "<ISO 8601>",
  "deploy_count": N,
  "status": "success|failed|unknown|halted",
  "workflow": "{config.deploy_workflow}",
  "branch": "{branch}",
  "commit_sha": "{HEAD short SHA at trigger time}",
  "run_id": "{run_id}",
  "run_url": "{run_url}",
  "health_check": "passed|failed|skipped",
  "health_retry_count": 0,
  "duration_seconds": N,
  "error_message": null
}

deploy_count: increment from previous value in the file (default 0 if file absent). commit_sha: captured via git rev-parse --short HEAD before triggering. duration_seconds: wall-clock time from trigger to final state write. error_message: null on success; short description on failure/unknown.

Print deployment summary:

run-deploy — <ISO timestamp>
Workflow : {config.deploy_workflow}
Branch   : {branch}
Commit   : {commit_sha}
Run ID   : {run_id}
Status   : success | failed | unknown | halted
Health   : passed | failed | skipped
Duration : {duration_seconds}s
Run URL  : {run_url}

If status == "failed": print Workflow failed. Investigate at {run_url} before retrying. If status == "halted": print HALT file appeared during deployment. Monitoring stopped. If health_check == "failed": print Post-deploy health check failed after retry. Manual rollback may be needed.


Graceful Degradation

ScenarioBehavior
HALT file presentPrint reason, exit immediately — no deployment
HALT file appears during monitorRecord status: "halted", skip health check, print warning
deploy_workflow not configuredPrint skip message, exit 0
Workflow file not found in .github/workflows/Print Workflow file {name} not found. Check config.deploy_workflow., exit non-zero
deploy_workflow_inputs invalid (non-object)Print deploy_workflow_inputs must be a key-value object., exit non-zero
gh not installedPrint install instructions, exit non-zero
Pre-deploy checklist failsPrint failing check, exit 0 (not an error — a safety gate)
Run ID not found after triggerRecord run_id: null, status "unknown", print warning
Monitor timeoutRecord status: "unknown", print manual check URL with configured timeout value
Workflow fails (non-zero exit)Record status: "failed", capture error_message from run logs, skip health check
Health check fails first attemptRetry once after deploy_health_wait_seconds, then record final result
health_endpoints missingRecord health_check: "skipped", note in report
service_health.json or test_coverage.json missingTreat as checklist failure for that check
deploy.json missingCreate fresh with deploy_count: 1

What ships with it

Read from the repository

Just SKILL.md. No reference files, no scripts.

Gives 0 of the 12 instructions most ship operate skills give in ~1.8k tokens

Counted across 1,077 of the 1,713 authors here whose files we hold, read 2026-09-06

  • Create GitHub releasein 44 of 1077, across 43 files
  • Run the test suitein 30 of 1077, across 25 files
  • Create and push git tagin 27 of 1077, across 26 files
  • Push commits and tagsin 27 of 1077
  • Create annotated tagin 25 of 1077, across 22 files
  • Ensure working tree is cleanin 24 of 1077
  • Check for product marketing context firstin 23 of 1077, across 6 files
  • Commit version bump changesin 22 of 1077, across 21 files
  • Update CHANGELOG.mdin 21 of 1077, across 20 files
  • Structure launch marketing across three channel typesin 20 of 1077, across 5 files
  • Commit and tag the releasein 20 of 1077, across 18 files
  • Update the CHANGELOG for new releasesin 19 of 1077

Said here and by no other author read

  • Check for HALT file before proceeding
  • Write deployment results to state file

Grouped from the skills themselves: near-identical wordings counted once, and counted by distinct author, so one author publishing three of these counts once. Length counted with cl100k_base; the agent that loads this file may tokenize it differently.

Keep looking

Skills are one crate of 611,178. Ordering is by how many stacks a row turns up in, so the top of any crate is what has actually been picked rather than what has the most stars.