Pr Writer

PR Writer

Write the PR body as a cover note for reviewers, not a changelog, template, validation log, or file-by-file summary.

Inspect the Change

Requires authenticated gh. Inspect the current branch, working tree, PR, base branch, commits, and full diff:

bashCopy code
git branch --show-current
git status --porcelain
gh pr view --json number,title,body,url,baseRefName,headRefName
gh repo view --json defaultBranchRef

If gh pr view reports that no PR exists, continue with first-time PR creation. For an existing PR, use its baseRefName; otherwise use the repository default branch. Set BASE, then inspect:

bashCopy code
git log "$BASE"..HEAD --oneline
git diff "$BASE"...HEAD

If on main or master, create a feature branch first. Ensure intended changes are committed and review the whole branch diff, not only the latest commit or existing PR text.

Core Rules

• Describe concrete changed behavior, affected surfaces, and reviewer impact before implementation detail.
• Explain motivation, risk, tradeoffs, migration, or review focus only when useful.
• Use the smallest structure that makes the change easier to review.
• Replace internal prompt or process terminology with specific behavior.
• When refreshing a PR, rewrite around the current full diff without narrating review history.

Titles

Use <type>(<scope>): <subject> or <type>: <subject>.

Allowed types: feat, fix, ref, perf, docs, test, build, ci, chore, style, meta, license, and revert.

• Describe the dominant full-branch change with the narrowest accurate type and scope.
• Use ! only when the change breaks an external contract, and explain the affected surface in the body.
• Avoid vague subjects such as update, cleanup, misc, fix stuff, or address feedback. Do not add a trailing period.
• Keep an existing title only when it still describes the whole diff.

Body Shape

Choose the minimum useful shape:

Default:

markdownCopy code
<What changed and what effect it has.>

<Why the approach, risk, migration, or review focus matters, if not obvious.>

For review-feedback updates, describe the resulting PR as a whole rather than the sequence of revisions.

Reviewer Aids

Use an aid only when it reduces reviewer reconstruction work:

• A compact before/after or interface example for changed contracts.
• A small Mermaid diagram for async flows or state transitions.
• A screenshot or recording note when visual evidence exists.
• A rollout, compatibility, risk, or review-order note when reviewers or adopters need it.

Introduce an artifact with one sentence explaining what reviewers should notice. Omit it when prose is clearer.

Boundaries

• Do not add default Summary, Changes, or Test Plan sections.
• Omit routine validation unless it changes risk assessment or explains meaningful regression coverage. For docs, skills, copy, or config changes, omit it by default.
• Do not paste commands, CI logs, validation dumps, commit logs, placeholders, or exhaustive file lists.
• Never include customer or organization names, user emails, support ticket contents, secrets, or PII.
• Use issue references only when verified from user input, branch names, commits, PR discussion, or tracker output. Fixes <issue> closes; Refs <issue> only links.

Create or Update

Create new PRs as drafts. Write the body to a temporary Markdown file, then run:

bashCopy code
gh pr create --draft --title '<title>' --body-file /tmp/pr-body.md

Update existing PRs with gh api:

bashCopy code
gh api -X PATCH repos/{owner}/{repo}/pulls/PR_NUMBER \
  -f title='<title>' \
  -F body=@/tmp/pr-body.md

Refresh the title and body when follow-up commits materially change scope, approach, breaking behavior, risk, migration, or review expectations. Skip typo-only, formatting-only, and rename-only follow-ups.

Examples

Small change:

markdownCopy code
The AI Customizations section now starts collapsed so it does not consume
sidebar space before users need it. Expanding it preserves the existing saved
preference behavior.

Breaking contract:

markdownCopy code
Run logs now emit chunk-level records instead of one skill-level record.
Consumers that read top-level `findings` must iterate over
`chunk.findings` for each record.

Before:

```json
{"skill": "security-review", "findings": [...]}
```

After:

```json
{"schemaVersion": 1, "chunk": {"index": 1, "findings": [...]}}
```