This skill handles the complete workflow of opening a pull request to add inheritance: true to a repository's .coderabbit.yaml file. It always operates via a personal fork to avoid needing direct push access to upstream repos.

When to Use This Skill

Use this skill as Step 6 of the /teams:coderabbit-inheritance-scanner command, after the user has confirmed they want to open PRs. Call it once per non-compliant repo that does not already have an open fix PR.

Do NOT call this skill in --dry-run mode. Instead, display the actions that would be taken for each repo.

Prerequisites

gh CLI authenticated with repo scope
workflow scope is needed if the upstream repo contains GitHub Actions workflow files (for gh repo sync). If missing, the skill will skip the repo with a warning.

Procedure

For each repo that needs a fix PR, run the following. The variables REPO (e.g., openshift/api), DEFAULT_BRANCH (e.g., master or main), and CFGFILE (e.g., .coderabbit.yaml) must be set.

# Derive names
REPO_NAME="${REPO##*/}"
GH_USER=$(gh api user --jq '.login')
WORKDIR="/tmp/cr-fix-workdir/${REPO_NAME}"
BRANCH_NAME="add-coderabbit-inheritance"

echo "Processing: ${REPO}"

# Step 1: Fork the repo (idempotent - safe to call if fork exists)
gh repo fork "${REPO}" --clone=false 2>&1 || true
sleep 2

# Step 2: Sync the fork's default branch with upstream
SYNC_OUTPUT=$(gh repo sync "${GH_USER}/${REPO_NAME}" --source "${REPO}" --branch "${DEFAULT_BRANCH}" 2>&1)
SYNC_EXIT=$?
if [ $SYNC_EXIT -ne 0 ]; then
  if echo "$SYNC_OUTPUT" | grep -q "workflow"; then
    echo "SKIP|${REPO}|Fork sync failed: workflow scope required. Run: gh auth refresh -h github.com -s workflow"
    # Continue to next repo in the calling loop
    return 1 2>/dev/null || continue
  fi
  echo "SKIP|${REPO}|Fork sync failed: ${SYNC_OUTPUT}"
  return 1 2>/dev/null || continue
fi

# Step 3: Delete any stale branch from a previous run
gh api -X DELETE "repos/${GH_USER}/${REPO_NAME}/git/refs/heads/${BRANCH_NAME}" 2>/dev/null || true
sleep 1

# Step 4: Clone the fork and create the fix branch
rm -rf "${WORKDIR}"
if ! gh repo clone "${GH_USER}/${REPO_NAME}" "${WORKDIR}" -- -b "${DEFAULT_BRANCH}" --depth 1 2>&1; then
  echo "ERROR|${REPO}|Clone failed"
  return 1 2>/dev/null || continue
fi
cd "${WORKDIR}" || { echo "ERROR|${REPO}|cd into workdir failed"; return 1 2>/dev/null || continue; }
git checkout -b "${BRANCH_NAME}" 2>&1 || { echo "ERROR|${REPO}|Branch creation failed"; cd /tmp; return 1 2>/dev/null || continue; }

# Step 5: Add inheritance: true to the config file
if [ ! -f "${CFGFILE}" ]; then
  echo "ERROR|${REPO}|Config file ${CFGFILE} not found in cloned repo"
  cd /tmp
  return 1 2>/dev/null || continue
fi

if ! grep -qE '^\s*inheritance:\s*(true|false)\b' "${CFGFILE}"; then
  # Prepend inheritance: true as the first line
  printf 'inheritance: true\n' | cat - "${CFGFILE}" > "${CFGFILE}.tmp" && mv "${CFGFILE}.tmp" "${CFGFILE}"
else
  # Replace existing inheritance: false with true
  sed -i.bak 's/^\(\s*\)inheritance:\s*false/\1inheritance: true/' "${CFGFILE}" && rm -f "${CFGFILE}.bak"
fi

# Step 6: Commit and push
git add "${CFGFILE}"
git commit -m "$(cat <<'COMMITEOF'
Add inheritance: true to .coderabbit.yaml

Ensures this repo inherits the org-wide CodeRabbit review rules
defined in https://github.com/openshift/coderabbit
COMMITEOF
)" 2>&1

git push origin "${BRANCH_NAME}" 2>&1

# Step 7: Create the PR from fork to upstream
PR_URL=$(gh pr create \
  --repo "${REPO}" \
  --head "${GH_USER}:${BRANCH_NAME}" \
  --base "${DEFAULT_BRANCH}" \
  --title "Add CodeRabbit inheritance for org-wide rules" \
  --body "$(cat <<'BODYEOF'
Adds `inheritance: true` to `.coderabbit.yaml` so this repo inherits the org-wide review rules from [openshift/coderabbit](https://github.com/openshift/coderabbit).

Without this setting, the repo's custom `.coderabbit.yaml` overrides the org-level configuration entirely, meaning org-wide review instructions and path-based rules are not applied.
BODYEOF
)" 2>&1)

cd /tmp

if echo "$PR_URL" | grep -q "https://github.com"; then
  echo "OK|${REPO}|${PR_URL}"
else
  echo "ERROR|${REPO}|PR creation failed: ${PR_URL}"
fi

sleep 1

Output Format

Pipe-separated fields per line:

STATUS|org/repo|detail

Where STATUS is one of:

OK - PR created successfully; detail is the PR URL
SKIP - Repo was skipped; detail is the reason (e.g., missing workflow scope)
ERROR - An error occurred; detail is the error message

Dry Run Output

When --dry-run is active, do NOT execute this skill. Instead, for each repo that would get a PR, display:

[DRY RUN] openshift/<repo>:
  1. Fork openshift/<repo> (or use existing fork)
  2. Sync fork's <default_branch> with upstream
  3. Create branch: add-coderabbit-inheritance
  4. Prepend "inheritance: true" to <config_file>
  5. Push branch to fork
  6. Open PR: "Add CodeRabbit inheritance for org-wide rules"
     From: <user>:add-coderabbit-inheritance -> openshift/<repo>:<default_branch>

Notes

The fork operation is idempotent - gh repo fork will not fail if the fork already exists.
The stale branch deletion (Step 3) ensures a clean state if the command was previously run and the PR was closed without merging.
macOS sed -i requires a backup extension; the .bak file is cleaned up.
The WORKDIR under /tmp/cr-fix-workdir/ is cleaned up before each clone to ensure fresh state.

coderabbit-inheritance-scanner-open-pr

Plus depuis ce dépôt

Plus depuis ce dépôt