
Merge Ready
- 46 installs
- 70 repo stars
- Updated July 26, 2026
- rysweet/amplihack
Helps with ai & agent building tasks.
About
merge-ready is a Claude Code skill for ai & agent building. It helps solo builders move faster with AI-assisted development.
- merge-ready
- AI & Agent Building
- AI-coding skill
Merge Ready by the numbers
- 46 all-time installs (skills.sh)
- +1 installs in the week ending Jul 26, 2026 (Skillselion tracking)
- Ranked #7,539 of 16,556 AI & Agent Building skills by installs in the Skillselion catalog
- Data as of Aug 2, 2026 (Skillselion catalog sync)
npx skills add https://github.com/rysweet/amplihack --skill merge-readyAdd your badge
Show developers this skill is listed on Skillselion. Paste this into your README.
| Installs | 46 |
|---|---|
| repo stars | ★ 70 |
| Last updated | July 26, 2026 |
| Repository | rysweet/amplihack ↗ |
What it does
Helps with ai & agent building tasks.
Files
Merge Ready
Use this skill as the final gate before calling a PR ready for review or merge.
This skill coordinates existing project workflows. It does not replace:
qa-teamfor outside-in scenario authoring and executionquality-auditfor iterative SEEK → VALIDATE → FIX reviewdefault-workflowfor substantive fixes discovered while making the PR merge-ready
Use pr-description-template.md to update the PR description with the required evidence.
Required outcome
A PR is merge-ready only when all of the following are true:
1. qa-team scenarios were written or updated, validated with gadugi-test validate, and actually run with gadugi-test run. 2. User-facing docs were updated when the change affects APIs, configuration, deployment, CLI behavior, or other external surfaces. 3. quality-audit completed at least 3 SEEK → VALIDATE → FIX cycles, continued past 3 if critical or high findings remained, and ended on a clean final cycle. 4. All GitHub Actions checks are green with 0 failures. 5. The PR description contains concrete evidence for criteria 1–4 and 6. 6. The diff contains no unrelated changes.
If any criterion is missing, the PR is not merge-ready.
Non-negotiable guardrails
- Do not treat "scenario YAML exists" as sufficient. The scenarios must be validated and run.
- Do not claim docs are irrelevant without checking the changed surfaces.
- Do not accept fewer than 3 quality-audit cycles.
- Do not accept a quality-audit result unless the final cycle is clean.
- Do not accept pending or failing CI.
- Do not mark a PR merge-ready until the PR description itself is updated with evidence.
- Do not silently ignore blockers such as missing environment access, missing PR access, or missing test tooling. Report them explicitly.
Workflow
1. Establish PR context
Identify the target PR from $ARGUMENTS if provided; otherwise use the current branch's PR.
Gather:
- PR number, URL, title, and current PR description
- changed files and current diff scope
- whether the change touches external or user-facing surfaces
- current CI status
Useful commands:
gh pr view "$PR" --json number,url,title,body
gh pr diff "$PR" --name-only
gh pr checks "$PR"
git diff --name-only origin/main...HEADIf no PR exists yet, you may still prepare evidence, but the final verdict must remain blocked until the PR exists and CI status is available.
2. Satisfy the QA criterion with qa-team
Invoke qa-team and ensure it covers the changed behavior from an external user perspective.
Required deliverables:
- scenario file path(s)
- proof that
gadugi-test validatesucceeded - proof that
gadugi-test runsucceeded - target environment used for the run (local or deployed)
- evidence location or execution output summary
Minimum commands:
gadugi-test validate path/to/scenario.yaml
gadugi-test run path/to/scenario.yamlIf the environment needed to run the scenario does not exist, stop and report a blocker. Do not downgrade this to "definition only".
3. Satisfy the docs criterion
Inspect the changed files and decide whether the PR changes any user-facing surface:
- APIs
- configuration
- deployment or operations flow
- CLI or TUI behavior
- user-visible workflows
If yes, update the corresponding docs and add links to those docs in the PR description.
If no, list the changed surfaces you checked and explain why each is internal-only. For example: "Changed surfaces reviewed: rust_runner.py, smart-orchestrator.yaml — both are internal orchestration plumbing with no CLI/API/config impact."
4. Satisfy the quality-audit criterion
Invoke quality-audit and require the full iterative loop:
- minimum 3 cycles
- each cycle is SEEK → VALIDATE (multi-agent consensus) → FIX
- continue past 3 cycles if critical or high findings remain
- final cycle must be clean: zero critical or high findings, and zero medium findings that pose correctness or security risks
Every fix uncovered during this audit must follow default-workflow. If the audit finds issues that require code or docs changes, switch to default-workflow for those fixes, complete the work, then return to this skill and resume verification.
Capture in the PR description:
- cycle count
- confirmed findings per cycle
- fixes applied
- convergence summary
- explicit statement that the final cycle was clean (zero critical/high findings, zero medium correctness/security findings)
5. Satisfy the CI criterion
Check all GitHub Actions status for the PR.
Rules:
- all checks must be green (skipped checks from conditional workflows are acceptable but must be listed)
- rerun only clearly flaky jobs
- fix real failures before continuing
- 0 failures is the only passing state
- if CI is pending, red, cancelled, or unavailable, the PR is not merge-ready
- document any skipped checks by name and skip reason in the PR description
Useful commands:
gh pr checks "$PR"
gh run list --branch "$(git branch --show-current)"If CI is pending, red, cancelled, or unavailable, the PR is not merge-ready.
6. Enforce scope
Review the diff and confirm the PR is focused on the intended change. If unrelated changes are present, report them as blockers and do not mark the PR ready.
7. Update the PR description
This step must not be performed until steps 2–6 have all completed and evidence has been collected. Do not update the PR description with placeholder or anticipated evidence.
Use pr-description-template.md and replace every placeholder with actual evidence. Keep the wording concise, but make the evidence concrete enough that a reviewer can verify the work without guessing.
If the PR description already has useful sections, merge the evidence into the existing structure instead of duplicating headings.
8. Return a binary verdict
End with one of these outcomes:
MERGE_READY: every criterion passed and the PR description was updatedNOT_MERGE_READY: one or more blockers remain
Use this format:
Verdict: MERGE_READY | NOT_MERGE_READY
Criteria:
- QA-team: pass | fail
- Docs: pass | fail | not-applicable
- Quality-audit: pass | fail
- CI: pass | fail
- Scope: pass | fail
- PR description evidence: pass | fail
Blockers:
- <concrete missing item or `none`>When to stop and hand back a blocker
Stop and report NOT_MERGE_READY when any of these are true:
- no runnable environment exists for the required gadugi scenario
- the PR or CI status cannot be accessed
- quality-audit has not converged to a clean final cycle
- docs changes are needed but not yet written
- the diff contains unrelated changes
Do not substitute promises or TODOs for evidence.
PR description evidence template
Copy these sections into the PR description and replace every placeholder with actual evidence.
Merge readiness
QA-team evidence
- Scenario files:
<path/to/scenario.yaml> - Validation command:
gadugi-test validate <scenario> - Validation result:
<passed / failed> - Run command:
gadugi-test run <scenario> - Run target:
<local env / deployed env> - Run result:
<passed / failed> - Evidence location:
<artifact path or log link>(must include pass/fail counts and command output, not just a claim)
Documentation
- User-facing docs impact:
<yes / no> - Updated docs:
<doc path or link, or "none"> - PR description links added:
<list of links or "n/a"> - Rationale if not applicable:
<list of changed surfaces checked and why each is internal-only>
Quality-audit
- Cycle 1 summary:
<findings, validation result, fixes> - Cycle 2 summary:
<findings, validation result, fixes> - Cycle 3 summary:
<findings, validation result, fixes> - Additional cycles:
<none or summary> - Final clean cycle:
<cycle number>(zero critical/high, zero medium correctness/security findings) - Fixes followed default-workflow:
<yes / no> - Convergence summary:
<why the audit is considered complete>
CI
- Checks command:
gh pr checks <pr> - Result:
<all green / failures remain / pending> - Skipped checks:
<list of skipped check names and skip reason, or "none"> - Flaky reruns performed:
<none or list> - Real failures fixed:
<none or summary>
Scope
- Changed files reviewed:
<summary or command used> - Unrelated changes:
<none or list>
Verdict
- Merge-ready:
<yes / no> - Remaining blockers:
<none or list>