drn/merge
agents/skills/merge/SKILL.md
Merge current branch to the default branch via GitHub PR merge. Use when ready to merge a PR or land a branch.
$ install --global
skillsauthnpx skillsauth add drn/dots mergeInstall this skill globally with one command. Works with Claude Code, Cursor, and Windsurf.
Security Scan Results
3 of 9 scanners reported clean
Some scanners were skipped, did not run, or reported a non-clean status. Review each row below.
3
Scanners Passed
9
Scanners in report
Clean
TrivyContainer and dependency vulnerability scanner
95%
Clean
SemgrepStatic code analysis for vulnerabilities
95%
Clean
mcp-scan (Snyk)Model Context Protocol security validation
95%
Skipped
Snyk (dep)Open source security scanning
50%
Skipped
Socket.devSupply chain security analysis
50%
Skipped
VirusTotalMulti-engine malware detection
50%
Skipped
CrowdStrikeAdvanced threat intelligence
50%
Skipped
OSV-ScannerOpen Source Vulnerability database check
50%
Skipped
OWASP Dep-Check
50%
Last scanned: Apr 26, 2026, 4:57 AM162.9s2 files scanned
SKILL.md
- name:
- merge
- allowed-tools:
- Bash(git add:*), Bash(git commit:*), Bash(git log:*), Bash(git diff:*), Bash(git status:*), Bash(git rebase:*), Bash(git checkout:*), Bash(git reset:*), Bash(bash ~/.claude/skills/merge/scripts/merge.sh:*), Bash(bash agents/skills/merge/scripts/merge.sh:*)
- description:
- Merge current branch to the default branch via GitHub PR merge. Use when ready to merge a PR or land a branch.
Context
- Current branch: !
git branch --show-current - Git status: !
git status --short - Remotes: !
git remote -v 2>/dev/null | head -10 - Base ref: !
git branch -r 2>/dev/null | grep -oE 'origin/(main|master)' | head -1 - Commits vs main (upstream): !
git log upstream/main..HEAD --oneline 2>/dev/null | head -50 - Commits vs master (upstream): !
git log upstream/master..HEAD --oneline 2>/dev/null | head -50 - Commits vs main (origin): !
git log origin/main..HEAD --oneline 2>/dev/null | head -50 - Commits vs master (origin): !
git log origin/master..HEAD --oneline 2>/dev/null | head -50 - Diff stat vs main (upstream): !
git diff upstream/main..HEAD --stat 2>/dev/null | head -50 - Diff stat vs master (upstream): !
git diff upstream/master..HEAD --stat 2>/dev/null | head -50 - Diff stat vs main (origin): !
git diff origin/main..HEAD --stat 2>/dev/null | head -50 - Diff stat vs master (origin): !
git diff origin/master..HEAD --stat 2>/dev/null | head -50
Phase Protocol
This skill participates in a phase chain. Read ~/.claude/skills/_shared/resources/phase-protocol.md for the full protocol.
Before merging: Read the latest .context/phases/ship-*.md for the PR number/URL if available.
After merge completes: Write a land-{ts}.md artifact to .context/phases/ (create with mkdir -p .context/phases). The Detail section should include the merge commit SHA and PR URL. Optionally archive all phase artifacts: mkdir -p .context/phases/archive && mv .context/phases/*.md .context/phases/archive/.
Your task
Merge the current branch into the default branch via GitHub PR merge. An existing PR is NOT required — one will be created if needed.
Step 0: Preflight — ahead/behind check
Before committing anything, verify the branch is in a mergeable state.
Run:
TARGET=$(git remote | grep -q '^upstream$' && echo upstream || echo origin)
git fetch "$TARGET" >/dev/null 2>&1 || true
DEFAULT_BRANCH=$(git branch -r 2>/dev/null | grep -oE "${TARGET}/(main|master)" | head -1 | sed "s|${TARGET}/||")
[ -z "$DEFAULT_BRANCH" ] && DEFAULT_BRANCH="master"
AHEAD=$(git log "$TARGET/$DEFAULT_BRANCH..HEAD" --oneline 2>/dev/null | wc -l | tr -d ' ')
BEHIND=$(git log "HEAD..$TARGET/$DEFAULT_BRANCH" --oneline 2>/dev/null | wc -l | tr -d ' ')
DIRTY=$(git status --short 2>/dev/null | wc -l | tr -d ' ')
echo "ahead=$AHEAD behind=$BEHIND dirty=$DIRTY default_branch=$DEFAULT_BRANCH"
Decide based on the output:
- ahead=0 and dirty=0 — Stop. Reply: "Nothing to merge — branch has no commits ahead of $DEFAULT_BRANCH and working tree is clean."
- Otherwise (ahead > 0, or dirty > 0) — Proceed to Step 1. The merge script handles fetch + rebase automatically, so being behind the default branch is fine.
Step 1: Commit uncommitted changes
If git status shows uncommitted changes, stage and commit them with an appropriate message. Skip if working tree is clean.
Step 2: Analyze commits and craft PR metadata
Review all commits and the full diff since the branch diverged from the default branch. Determine:
- Title: Concise imperative summary of what the branch accomplishes (not individual commits). Under 72 characters.
- Body: Short description of the changes. Do NOT include Co-Authored-By — the script adds it.
If there are no prior commits (only what was committed in step 1), base the summary on that commit.
Step 3: Run the merge script
Resolve the script path — use the first that exists:
~/.claude/skills/merge/scripts/merge.sh(deployed via symlink)agents/skills/merge/scripts/merge.sh(repo-relative, for development/workspaces)
bash <script-path> [--method <squash|merge|rebase>] "<title>" "<body>"
Choosing a merge method
The script defaults to squash (collapses all commits into one). Pick a different method when squash would lose intentional structure:
--method squash(default) — single commit, ideal for a stream of WIP commits.--method merge(alias--merge) — preserves all commits via a merge commit. Use when each commit is independently meaningful and you want to keep the merge boundary visible on master.--method rebase(alias--rebase) — preserves all commits linearly via fast-forward / rebase. Use this when the user has deliberately split work into multiple commits (e.g., the user ran/squashto condense 7 commits into 2 separate logical commits, and wants both on master). A naive squash merge here would silently undo that intent.
Heuristic: if the branch has more than one commit AND the commit messages look distinct rather than incremental ("WIP", "fix typo"), confirm with the user before using the default squash. The deliberate-split case is common after /squash.
Handle the exit code
-
Exit 0 — Show the output block verbatim as your final response. Do not add commentary.
-
Exit 2 (rebase conflict) — Resolve conflicts:
- Read the conflicting files from stderr
- Open each file, resolve the conflict
git addresolved filesgit rebase --continue- Repeat if more conflicts
- Re-run:
bash <script-path> --skip-rebase [--method <method>] "<title>" "<body>" - Show the output block verbatim
Special case: "distinct types" conflicts. If the rebase reports conflicts because the branch converted regular files to symlinks (or vice versa) while master independently edited those same files, do NOT naively resolve with
--theirsor--ours. Both sides "win" — the branch's symlink target was sourced from old master content, so keeping the branch's version silently loses master's content updates.Recovery pattern:
git rebase --abort.- Drop the symlink-conversion commit from the branch (e.g.,
git rebase -i masterand drop it, orgit reset --soft <sha-before-symlink-commit>then re-stage selectively). - Rebase the remaining commits onto master cleanly.
- Rebuild the symlink-conversion commit fresh against current master — re-source the symlink target's content from the current master file, not from the original AGENTS.md content captured on the branch.
- Re-run the merge script with
--skip-rebase.
-
Exit 3 — Tell the user: "Nothing to merge — branch has no commits ahead of the default branch."
-
Exit 4 (review blocked) — The PR requires review and auto-merge is not available. Tell the user: "PR requires an approving review before it can merge. Auto-merge is not enabled on this repository — ask a reviewer to approve, then re-run /merge."
-
Exit 1 — Report the error from stderr.
Related Skills
drn/address-comments
development
VerifiedTrustedCommunity
Walk every unresolved review thread on a PR, triage each one, reply with a rationale of whether or not the comment will be acted upon, make the code change if warranted, and mark the thread resolved. Use when the user asks to address only the open PR comments without re-running CI, respond to review feedback, resolve review threads, or clear bot comments on a PR.
23SKILL.mdUpdated May 16, 2026
drn/rereview-loop
tools
VerifiedTrustedCommunity
Iteratively run /rereview, fix the findings, and loop until reviewers approve clean. Use for iterative automated review, when you want /rereview to loop until clean, or for a paranoid pre-merge review that auto-addresses every blocker.
23SKILL.mdUpdated May 13, 2026
drn/plannotator-visual-explainer
development
VerifiedTrustedCommunity
Generate self-contained HTML visualizations with Plannotator theming. Use for implementation plans, PR explainers, architecture diagrams, data tables, slide decks, and any visual explanation of technical concepts. Plans and PR explainers follow Plannotator's prescriptive approach; all other visual content delegates to nicobailon/visual-explainer.
23SKILL.mdUpdated May 13, 2026
drn/plannotator-setup-goal
development
VerifiedTrustedCommunity
Create reviewed Codex goal setup packages for long-running /goal work. Use when the user wants to turn an idea, backlog, project mission, or vague objective into durable goal files under a project goals slug folder, with Plannotator review gates for brief, narrative plan with acceptance criteria, verification, blockers, and the final /goal prompt.
23SKILL.mdUpdated May 13, 2026