| name | generate-pr |
| description | Generate a complete pull request from the branch diff and open or update it. Use when the user asks to create/open a PR, fill the PR template, or mentions "/generate-pr". Two modes by where you stand: from a feature branch it targets `dev` (default); from `dev` it targets `main` (promotion PR). Fills EVERY section of .github/pull_request_template.md grounded in the diff, and runs `gh pr create`/`gh pr edit`. Honors DRY_RUN=1 to emit out/pr-body.md instead of touching GitHub. |
| license | MIT |
| allowed-tools | Bash, Read, Write |
Generate a Pull Request from the Branch Diff
Overview
Produce a fully-filled pull request — every applicable section of
the project's .github/pull_request_template.md, every checkbox the
evidence supports — and open or update the PR. Invocation is the authorization: run end-to-end
without a confirm prompt.
Core principle — deterministic vs voice (the constitution's A6). You (the agent executing this skill)
are the voice; there is no separate model call. The split is:
- Deterministic ("what is true"): what changed, which type/area boxes to tick, which refs actually
appear, whether a secret leaked. Gathered from
git/gh — never guessed.
- Voice (you): the Summary, the Changes bullets, the verification narrative — prose grounded only
in the deterministic facts.
Never invent. A reference, statistic, ticket id, or sign-off that is not in the diff/commits does not
go in the PR. Under-filling a box is recoverable; a fabricated claim is not (provenance: the constitution's
A15). When a fact has no evidence, write n/a.
Arguments & modes
Two modes, decided by the branch you are standing on:
- Work PR (from a feature branch) — target defaults to
dev (work lands on dev).
- Promotion PR (from
dev) — target defaults to main (proven-good work goes to prod). This is the
release path: main only moves via a PR from dev.
Options:
$1 (optional) — explicit target branch, overriding the default for the mode.
DRY_RUN=1 (env) — emit the body to out/pr-body.md and print the gh command instead of touching
GitHub. Also the automatic fallback when gh is missing/unauthenticated. This is how you test the
skill without spamming real PRs.
Step 0 — Preconditions (deterministic)
CURRENT=$(git rev-parse --abbrev-ref HEAD)
case "$CURRENT" in
main|master) echo "On $CURRENT — prod has nothing to PR. Checkout dev or a feature branch." ; exit 1 ;;
dev) TARGET="${1:-main}" ;;
*) TARGET="${1:-dev}" ;;
esac
if [ "$TARGET" = "main" ] && [ "$CURRENT" != "dev" ]; then
echo "Only dev may PR into main (got: $CURRENT). Target dev instead." ; exit 1
fi
git rev-parse --verify --quiet "$TARGET" >/dev/null || { echo "Target branch '$TARGET' not found." ; exit 1; }
git diff --stat "$TARGET...HEAD"
git diff --name-only "$TARGET...HEAD"
git log "$TARGET..HEAD" --format='%s%n%b%n---'
- Empty diff (
git diff --quiet "$TARGET...HEAD") ⇒ stop: "No changes vs $TARGET; nothing to PR."
- Capture the full
git diff "$TARGET...HEAD" to read the actual hunks. Size guard: if it is
larger than ~1500 lines, do not read every line — ground your prose in the commit subjects/bodies +
--stat + a few sampled hunks from the most-changed files.
Step 1 — Deterministic classification ("what is true")
Type of change — parse Conventional-Commit types from the git log subjects and tick every type present:
| Subject prefix | Box |
|---|
feat | feat — new capability |
fix | fix — bug fix |
docs | docs — documentation only |
refactor | refactor — no behavior change |
perf | perf — performance |
test | test — adds/updates tests |
chore, build, ci, style | chore — tooling/build/deps |
Area touched — map the changed paths (git diff --name-only) to whatever area checkboxes the project's
template defines (read it in Step 3). Tick each area the diff actually hits; if the template has no area
section, skip this. Ground the mapping in real paths, never a guess.
Traceability refs — extract, never invent. Search the commit messages and the diff text for the
id schemes this project uses (axioms A#, decisions D#, plan tasks, issues):
LOG=$(git log "$TARGET..HEAD" --format='%B')
DIFF=$(git diff "$TARGET...HEAD")
printf '%s\n%s\n' "$LOG" "$DIFF" | grep -oE '\b(A|P|D)[0-9]+\b' | sort -u
printf '%s\n' "$LOG" | grep -oiE '(closes|fixes|refs) #[0-9]+'
Fill Motivation & context from these. If a field has no hit, write n/a — do not guess a
plausible id.
Secret / PII scan (informs the "No secrets/PII" box; also a hard stop):
git diff --name-only "$TARGET...HEAD" | grep -qxE '\.env(\..*)?' && { echo "ABORT: .env in diff"; exit 1; }
git diff "$TARGET...HEAD" | grep -nE '(sk-[A-Za-z0-9]{12,}|API_KEY *= *["'\'']?[A-Za-z0-9]|-----BEGIN [A-Z]+ PRIVATE KEY-----)' && \
{ echo "ABORT: a secret-looking value is in the diff — do not open a PR."; exit 1; }
Step 2 — Prose ("voice" — you write it, grounded only in Step 0/1 facts)
- Title — Conventional-Commit aggregate: the dominant
type(scope) across the branch + a concise
subject, e.g. feat(auth): add OAuth device flow. Lowercase type, imperative subject.
- Summary — 1–3 sentences: what this PR does and why.
- Changes — grounded bullets, one per logical file-group from
--stat/--name-only. Specific, not
vague ("add render.ts multi/single/compact modes", not "improve rendering").
- How it was verified — honest. List the relevant commands from the project's
AGENTS.md §Commands
for the touched areas, plus any tests the branch added. Do not claim a command passed unless you
actually ran it and saw it pass (verification-before-completion / the constitution's A2). If nothing
was run, say which commands a reviewer should run.
Step 3 — Assemble the body
Read the template fresh so the body never drifts from the source of truth:
cat .github/pull_request_template.md
Every genesis repo ships this file (Phase 1 lays it down from templates/.github/). If it is absent (a
non-genesis repo), fall back to a built-in default section set in this order: Summary · Type of change · Area
touched · Changes · Motivation & context · How it was verified · Definition of Done · Human sign-off — then
proceed exactly as below. Reproduce its structure verbatim — same headers, same order, keep any <details> blocks — replacing
each <!-- … --> placeholder with filled content. Strip the leading HTML instruction comment. Delete a
whole section only if it genuinely does not apply (the template usually says so). Write the result to a
temp file for gh:
BODY=$(mktemp); printf '%s\n' "$ASSEMBLED_BODY" > "$BODY"
Step 4 — Definition-of-Done checkboxes
Tick a box only when the evidence supports it; otherwise leave it unchecked. An honest blank beats a
fabricated tick. Common ones:
- Reproducible — tick if the diff introduced no
Date.now()/Math.random() (or equivalent) into core
logic: git diff "$TARGET...HEAD" -- 'src/**' | grep -nE 'Date\.now\(|Math\.random\(' (no hits ⇒ safe).
- No secrets / PII — tick from the Step 1 scan (it passed).
- Tests — tick only if the branch added/updated tests AND you ran them green.
- Commits Conventional + trailer + correct target — verify the
Co-Authored-By trailer is present
(git log "$TARGET..HEAD" --format='%b' | grep -q 'Co-Authored-By:') and the target matches the mode:
dev from a feature branch, main only from dev.
Human-sign-off blocks (if the template has any — e.g. "reviewed by domain expert") — never auto-tick
a box that asserts a human's judgment. Leave it blank and add one line under the checklist noting the
sign-off is pending the operator's review. (This generalizes the constitution's gate rule, A21.)
Step 5 — Create or update the PR (fully automated)
if [ -n "$DRY_RUN" ] || ! command -v gh >/dev/null || ! gh auth status >/dev/null 2>&1; then
mkdir -p out; cp "$BODY" out/pr-body.md
echo "Wrote out/pr-body.md. To open the PR:"
echo "gh pr create --base $TARGET --head $CURRENT --title \"$TITLE\" --body-file out/pr-body.md"
exit 0
fi
if NUM=$(gh pr view --json number -q .number 2>/dev/null); then
gh pr edit "$NUM" --title "$TITLE" --body-file "$BODY"
gh pr view "$NUM" --json url -q .url
else
gh pr create --base "$TARGET" --head "$CURRENT" --title "$TITLE" --body-file "$BODY"
fi
Print the resulting PR URL.
Hard rules (red flags — stop if you catch yourself doing these)
- Writing an
A##/D##/#issue ref that is not in the commits or diff. → Use n/a.
- Ticking a human-sign-off box without the human. → Leave blank, note it pending.
- Claiming a verification command passed that you did not run. → Say what a reviewer should run instead.
- Hardcoding/paraphrasing the template instead of reading
.github/pull_request_template.md. → Read it fresh.
- Pasting any
.env value or secret into the body. → Already aborted in Step 1; never reach here.
- Opening against
main from anything other than dev. → Features PR into dev; only dev promotes to main.