prepare-pr-summary

name	prepare-pr-summary
metadata	{"author":"rhoai3-coding-demo","version":"1.1.0","platform-family":"rhoai","platform-baseline":"repo","ocp-baseline":"repo","skill-group":"Project Structure"}
description	Prepare a pull request summary following the project template. Use before opening a PR to generate the summary, risk assessment, rollback plan, and AI disclosure. Reads the git diff and produces structured output matching the PR format defined in AGENTS.md. Do NOT use for making code changes (use other skills/rules), reviewing manifests (use review-gitops-change), or validating steps (use validate-demo-step).

Prepare PR Summary

Use this skill to generate a complete PR summary before opening a pull request.

When to invoke

• Before running gh pr create
• When the user asks to "prepare a PR" or "write a PR summary"
• After completing a set of changes that will be submitted as a PR

Inputs needed

• The current branch diff (git diff main...HEAD or similar)
• List of commits on the branch
• Whether validation was run against a live cluster
• Which AI tools were used during the work

Workflow

1. Run git diff main...HEAD --stat to understand scope.
2. Run git log main..HEAD --oneline to see commit history.
3. Read the changed files to understand the purpose.
4. Identify security-sensitive changes (RBAC, gateway, auth, secrets, MaaS).
5. Check if docs were updated alongside behavior changes.
6. Determine risk level.
7. Write rollback instructions.
8. Produce the PR summary.

Output format

Use the following structure (from AGENTS.md):

## Summary

[1-3 sentences describing what changed]

## Why

[Why this change is needed — the problem or improvement]

## Changed files

- [file path]: [what changed]
- [file path]: [what changed]

## Validation

Commands run:

\`\`\`bash
[actual commands run]
\`\`\`

Result:

\`\`\`text
[actual output or summary]
\`\`\`

[If not validated: "Not validated against a live OpenShift cluster. Static review only."]

## Risk

- [x] Low / Medium / High

Risk notes: [explain what could go wrong]

## Rollback

[How to revert: git revert, Argo CD sync, manual steps]

## Security and governance checklist

- [x] No secrets or credentials committed
- [x] No unintended RBAC expansion
- [x] No unintended model access change
- [x] No direct bypass of MaaS unless explicitly documented
- [x] Trust boundary language remains accurate
- [x] Docs updated where needed

## AI assistance

- [x] AI assistance used

- Tool/model: [e.g., Cursor with Claude]
- Scope of assistance: [e.g., planning, code edits, documentation]
- Human review performed: [e.g., full diff reviewed by <name>]
- Validation performed: [e.g., bash -n, kustomize build, validate.sh]

Risk assessment guidelines

Risk level	When to use
Low	Docs-only, typo fixes, comment changes, non-functional additions
Medium	New manifests, script changes, configuration changes with validation
High	RBAC changes, gateway/auth policy changes, model access changes, workaround removal, operator version changes

What this skill must never do

• Claim validation that was not performed
• Omit AI assistance disclosure
• Mark security checklist items as checked without verification
• Produce a summary that doesn't match the actual diff