| name | titan-close |
| description | Split branch commits into focused PRs, compile issue tracker, generate final report with before/after metrics (Titan Paradigm Phase 5) |
| argument-hint | <--dry-run to preview without creating PRs> |
| allowed-tools | Bash, Read, Write, Glob, Grep, Edit |
Titan CLOSE — PR Splitting & Final Report
You are running the CLOSE phase of the Titan Paradigm.
Your goal: analyze all commits on the current branch, split them into focused PRs for easier review, compile the issue tracker from all phases, capture final metrics, and generate a comprehensive audit report.
Context budget: This phase reads artifacts and git history. Keep codegraph queries targeted — only for final metrics comparison.
Dry-run mode: If $ARGUMENTS contains --dry-run, preview the PR split plan and report without creating PRs or pushing branches.
Step 0 — Pre-flight: find and consolidate the Titan session
-
Locate the Titan session. All prior phases (RECON → GAUNTLET → SYNC → GATE) may have run across different worktrees or branches. You need to consolidate their work.
git worktree list
For each worktree, check for Titan artifacts:
ls <worktree-path>/.codegraph/titan/titan-state.json 2>/dev/null
Also check branches (including remote):
git branch -a --list '*titan*'
git branch -a --list '*refactor/*'
Decision logic:
- Found exactly one worktree/branch with
titan-state.json: Read its currentPhase. If it's "sync" or later, this is the right session. Merge its branch into your worktree.
- Found a worktree but
currentPhase is earlier than expected (e.g., "recon" or "gauntlet"): The pipeline may not be complete. Keep searching — there may be a more advanced worktree. If nothing better found, ask the user: "Found Titan state at <path> with phase <phase>. The pipeline appears incomplete. Continue anyway, or should I look elsewhere?"
- Found multiple worktrees with
titan-state.json: List them all with currentPhase, lastUpdated, and branch name. The Titan pipeline may have been split across worktrees (RECON in one, GAUNTLET in another). Merge them in phase order into your worktree. If there's ambiguity (e.g., two worktrees at the same phase), ask the user.
- Found branches but no worktrees: Merge the titan branch(es) in phase order:
git merge <branch> --no-edit
- Found nothing: Stop: "No Titan session found in any worktree or branch. Run
/titan-recon first."
-
Ensure worktree isolation:
git rev-parse --show-toplevel && git worktree list
If not in a worktree, stop: "Run /worktree first."
-
Sync with main:
git fetch origin main && git merge origin/main --no-edit
If there are merge conflicts, stop: "Merge conflict detected. Resolve conflicts and re-run /titan-close."
-
Load artifacts. Read:
.codegraph/titan/titan-state.json — session state, baseline metrics, progress
.codegraph/titan/GLOBAL_ARCH.md — architecture document
.codegraph/titan/gauntlet.ndjson — full per-target audit data (pillar verdicts, metrics, violations)
.codegraph/titan/gauntlet-summary.json — audit result totals
.codegraph/titan/sync.json — execution plan (commit grouping)
.codegraph/titan/gate-log.ndjson — validation history (may not exist if gate wasn't run)
.codegraph/titan/issues.ndjson — issue tracker from all phases
.codegraph/titan/arch-snapshot.json — pre-forge architectural snapshot (communities, structure, drift). Use for before/after comparison in the Metrics section. May not exist if capture failed.
.codegraph/titan/drift-report.json — cumulative drift reports from all phases. May not exist if no drift was detected.
.codegraph/titan/grind-targets.ndjson — grind phase adoption targets and outcomes. Each line: {target, file, phase, classification, reason, consumers, pattern, timestamp}. May not exist if grind wasn't run.
If titan-state.json is missing after the search, stop: "No Titan session found. Run /titan-recon first."
When called from /titan-run: The orchestrator already ensured worktree isolation, synced with main, and all artifacts are in the current worktree. Steps 0.1–0.3 (worktree search, isolation check, main sync) can be skipped if the orchestrator tells you to skip them.
-
Detect version. Extract from package.json:
node -e "console.log(require('./package.json').version)"
Step 1 — Drift detection: final staleness assessment
CLOSE is the last phase — it must assess the full pipeline's freshness before generating the report.
-
Compare main SHA:
git rev-parse origin/main
Compare against titan-state.json → mainSHA.
-
If main has advanced, calculate full drift:
git rev-list --count <mainSHA>..origin/main
git diff --name-only <mainSHA>..origin/main
-
Read all prior drift reports from .codegraph/titan/drift-report.json (a JSON array of entries, one per phase that detected drift). This shows the cumulative drift across the pipeline.
-
Assess overall pipeline freshness:
| Level | Condition | Action |
|---|
| fresh | mainSHA matches current main, no drift reports with severity > low | Generate report normally |
| acceptable | Some drift detected but phases handled it (re-audited stale targets) | Generate report — note drift in Executive Summary |
| stale | Significant unaddressed drift: >10 commits behind, >20% of audited targets changed on main since audit | Warn user: "Pipeline results are partially stale. N targets were modified on main after being audited. The report will flag these. Consider re-running /titan-gauntlet for affected targets before finalizing." |
| expired | >50 commits behind OR >50% of targets changed OR architecture-level changes (new directories in src/) | Stop: "Pipeline results are too stale to produce a reliable report. Run /titan-recon for a fresh baseline." |
-
Write final drift assessment to the drift report (same schema, "detectedBy": "close").
-
Include drift summary in the report. The final report's Executive Summary and Recommendations sections must reflect any staleness. Stale targets should be called out in a "Staleness Warnings" subsection.
Step 2 — Collect branch commit history
git log main..HEAD --oneline --no-merges
git log main..HEAD --format="%H %s" --no-merges
Extract: total commit count, commit messages, SHAs. If zero commits, stop: "No commits on this branch. Nothing to close."
For each commit, get the files changed:
git diff-tree --no-commit-id --name-only -r <sha>
Step 3 — Classify commits into PR groups
Analyze commit messages and changed files to group commits into focused PRs. Each PR should address a single concern for easier review.
Grouping strategy (in priority order)
Use sync.json execution phases as the primary guide if available:
- Dead code cleanup — commits removing dead symbols
- PR title:
chore: remove dead code identified by Titan audit
- Shared abstractions — commits extracting interfaces/utilities
- PR title:
refactor: extract <abstraction> from <source>
- Cycle breaks — commits resolving circular dependencies
- PR title:
refactor: break circular dependency in <domain>
- Decompositions — commits splitting complex functions/files
- PR title:
refactor: decompose <target> in <domain>
- Quality fixes — commits addressing fail-level violations
- Group by domain:
fix: address quality issues in <domain>
- Warning improvements — commits addressing warn-level issues
- Group by domain:
refactor: improve code quality in <domain>
- Helper adoption — commits from grind phase adopting dead helpers into callsites
- PR title:
refactor: adopt dead helpers in <domain>
- Look for
grind(...) commit prefixes or commits touching files listed in grind-targets.ndjson
Fallback grouping (if no sync.json)
Group by changed file paths — commits touching the same directory/domain go together. Use commit message prefixes (fix:, refactor:, chore:) as secondary signals.
Rules for grouping
- A PR should touch one domain where possible
- A PR should address one concern (don't mix dead code removal with refactors)
- Order PRs so dependencies come first (if PR B depends on PR A's changes, A merges first)
- Each PR must be independently reviewable — no PR should break the build alone
- If a commit touches files across multiple concerns, assign it to the primary concern and note the cross-cutting nature in the PR description
Record the grouping plan:
[
{
"pr": 1,
"title": "...",
"concern": "dead_code|abstraction|cycle_break|decomposition|quality_fix|warning|adoption",
"domain": "<domain name>",
"commits": ["<sha1>", "<sha2>"],
"files": ["<file1>", "<file2>"],
"dependsOn": [],
"description": "..."
}
]
Step 4 — Capture final metrics
Rebuild the graph and collect current metrics:
codegraph build
codegraph stats --json
codegraph complexity --health --above-threshold -T --json --limit 50
codegraph roles --role dead -T --json
codegraph roles --role core -T --json
codegraph cycles --json
Extract: totalNodes, totalEdges, totalFiles, qualityScore, functions above threshold, dead symbol count, core symbol count, cycle count.
Also get the worst offenders for comparison:
codegraph complexity --health --sort effort -T --json --limit 10
codegraph complexity --health --sort bugs -T --json --limit 10
codegraph complexity --health --sort mi -T --json --limit 10
Architecture comparison (if arch-snapshot.json exists)
If .codegraph/titan/arch-snapshot.json was captured before forge, compare its structure data against current codegraph structure --depth 2 --json output. Report cohesion changes per directory (improved / degraded / unchanged). Include in the "Metrics: Before & After" section of the report.
Grind metrics (if grind-targets.ndjson exists)
If .codegraph/titan/grind-targets.ndjson exists, parse it and cross-reference with titan-state.json → grind:
- Targets processed: count entries in
grind.processedTargets (from titan-state.json)
- Targets failed: count entries in
grind.failedTargets (from titan-state.json)
- False positives identified: count entries with
classification: "false-positive" in grind-targets.ndjson (or from grind.falsePositives in titan-state.json)
- Adopted: count entries with
classification: "adopt" in grind-targets.ndjson
- Dead symbol delta: from
grind.deadSymbolDelta in titan-state.json (or compute from baseline vs final dead symbol counts)
Include these in the Metrics: Before & After section and the Grind Results report section.
Compute deltas
Compare final metrics against titan-state.json baseline:
| Metric | Baseline | Final | Delta |
|---|
| Quality Score | from state | from stats | +/- |
| Functions above threshold | from state | from complexity | +/- |
| Dead symbols | from state | from roles | +/- |
| Cycles | from state | from cycles | +/- |
| Total nodes | from state | from stats | +/- |
Step 5 — Compile the issue tracker and open GitHub issues
Read .codegraph/titan/issues.ndjson. Each line is a JSON object:
{"phase": "recon|gauntlet|sync|gate", "timestamp": "ISO 8601", "severity": "bug|limitation|suggestion", "category": "codegraph|tooling|process|codebase", "description": "...", "context": "optional detail"}
Group issues by category and severity. Summarize:
- Codegraph bugs: issues with codegraph itself (wrong output, crashes, missing features)
- Tooling issues: problems with the Titan pipeline or other tools
- Process notes: suggestions for improving the Titan workflow
- Codebase observations: structural concerns beyond what the audit covered
5b. Open GitHub issues
Pre-check: Verify gh is available and authenticated before attempting issue creation:
gh auth status 2>&1 || echo "GH_UNAVAILABLE"
If GH_UNAVAILABLE, skip issue creation entirely and note in the report: "GitHub issues were not created — gh CLI is not available or not authenticated. Create them manually from the Issues section below."
For each issue with severity bug or limitation, create a GitHub issue using gh:
BODY=$(mktemp)
cat > "$BODY" <<'ISSUE_BODY'
Discovered during Titan audit (phase: <phase>, date: <timestamp>).
<description>
<context field, if present>
- **Titan phase:** <phase>
- **Severity:** <severity>
- **Category:** <category>
ISSUE_BODY
gh issue create --title "<category>: <short description>" --body-file "$BODY" --label "titan-audit"
rm -f "$BODY"
Using --body-file with a temp file avoids quoting/expansion issues that can arise when issue descriptions contain backticks, $() sequences, or literal EOF strings.
Rules for issue creation:
- Only open issues for
bug and limitation severity. Suggestions and observations go in the report only — they are not actionable enough for standalone issues.
- Check for duplicates first: Run
gh issue list --search "<short description>" --state open --limit 5 before creating. If a matching open issue exists, skip it and note "existing issue #N" in the report.
- Label: Use
titan-audit label. If the label doesn't exist, create it: gh label create titan-audit --description "Issues discovered during Titan audit" --color "d4c5f9" 2>/dev/null || true
- Record each created issue number for inclusion in the report's Issues section.
For suggestion severity entries and entries with category: "codebase", include them in the report's Issues section but do NOT create GitHub issues.
Step 6 — Compile the gate log
Read .codegraph/titan/gate-log.ndjson. Summarize:
- Total gate runs
- Pass / Warn / Fail counts
- Rollbacks triggered
- Most common failure reasons
Step 7 — Generate the report
Record CLOSE completion timestamp
Before writing the report, record phaseTimestamps.close.completedAt so the Pipeline Timeline has accurate data for the CLOSE row. (titan-run also records this after titan-close returns as a safety backstop, but by then the report is already written.)
node -e "const fs=require('fs');const s=JSON.parse(fs.readFileSync('.codegraph/titan/titan-state.json','utf8'));s.phaseTimestamps=s.phaseTimestamps||{};s.phaseTimestamps['close']=s.phaseTimestamps['close']||{};s.phaseTimestamps['close'].completedAt=new Date().toISOString();fs.writeFileSync('.codegraph/titan/titan-state.json',JSON.stringify(s,null,2));"
Report path
generated/titan/titan-report-v<version>-<date>T<time>.md
Where:
<version> is the package.json version (e.g., 3.1.5)
<date> is YYYY-MM-DD
<time> is HH-MM-SS (local time, hyphen-separated for filesystem safety)
Example: generated/titan/titan-report-v3.1.5-2026-03-17T14-30-00.md
mkdir -p generated/titan
Report structure
Write the report as Markdown:
# Titan Audit Report
**Version:** <package version>
**Date:** <start date> → <end date>
**Branch:** <branch name>
**Target:** <resolved path>
---
## Executive Summary
<2-3 sentences: what was audited, key outcomes, overall health change>
---
## Pipeline Timeline
Read `titan-state.json → phaseTimestamps` for real wall-clock data. If `phaseTimestamps` exists, use the recorded ISO 8601 timestamps to compute durations. If it does not exist (older pipeline run), derive timing from git commit timestamps as a fallback — **never invent or guess timestamps.**
**Duration computation:** For each phase with `startedAt` and `completedAt`, compute duration as the difference in minutes/hours. For forge, also note the first and last commit timestamps from `git log`.
| Phase | Duration | Notes |
|-------|----------|-------|
| RECON | <computed from phaseTimestamps.recon> | — |
| GAUNTLET | <computed from phaseTimestamps.gauntlet> | <iterations count if resuming> |
| SYNC | <computed from phaseTimestamps.sync> | — |
| FORGE | <computed from phaseTimestamps.forge> | <commit count>, first at <time>, last at <time> |
| GRIND | <computed from phaseTimestamps.grind> | <targets processed>, <adoptions made> |
| GATE | across forge/grind | <total runs> inline with forge/grind commits |
| CLOSE | <computed from phaseTimestamps.close> | — |
| **Total** | <sum of all phases> | — |
**If `phaseTimestamps` is missing:** Fall back to git log timestamps. Use the earliest and latest commit timestamps from `git log main..HEAD --format="%ai"` to bound the forge phase. For analysis phases (recon, gauntlet, sync), use `titan-state.json → initialized` and `lastUpdated` as rough bounds. Mark the durations as "~approximate" in the table.
---
## Metrics: Before & After
| Metric | Baseline | Final | Delta | Trend |
|--------|----------|-------|-------|-------|
| Quality Score | X | Y | +/-Z | arrow |
| Total Files | ... | ... | ... | ... |
| Total Symbols | ... | ... | ... | ... |
| Functions Above Threshold | ... | ... | ... | ... |
| Dead Symbols | ... | ... | ... | ... |
| Cycle Count | ... | ... | ... | ... |
| Avg Halstead Bugs | ... | ... | ... | ... |
| Avg Maintainability Index | ... | ... | ... | ... |
### Complexity Improvement: Top Movers
<Table of functions that improved the most — dropped below thresholds or reduced halstead.bugs>
### Remaining Hot Spots
<Table of functions still above thresholds — carried forward for next Titan run>
---
## Audit Results Summary
**Targets audited:** <N>
**Pass:** <N> | **Warn:** <N> | **Fail:** <N> | **Decompose:** <N>
### By Pillar
| Pillar | Pass | Warn | Fail |
|--------|------|------|------|
| I — Structural Purity | ... | ... | ... |
| II — Data & Type Sovereignty | ... | ... | ... |
| III — Ecosystem Synergy | ... | ... | ... |
| IV — Quality Vigil | ... | ... | ... |
### Most Common Violations
<Top 5 violation types with counts>
---
## Grind Results (if grind ran)
**Targets processed:** <N> | **Adopted:** <N> | **Failed:** <N> | **False positives:** <N>
### Adoption Summary
<Table: target name, dead symbols before, dead symbols after, delta, files modified>
### False Positives Logged
<Table: target name, reason (dynamic import / re-export / closure-local / type noise), logged to issues.ndjson>
> Omit this section entirely if `grind-targets.ndjson` does not exist.
---
## Changes Made
### Commits: <total count>
<Table: SHA (short), message, files changed, domain>
### PR Split Plan
| PR # | Title | Concern | Domain | Commits | Files | Depends On |
|------|-------|---------|--------|---------|-------|------------|
| 1 | ... | ... | ... | N | N | — |
| 2 | ... | ... | ... | N | N | PR #1 |
---
## Gate Validation History
**Total runs:** <N>
**Pass:** <N> | **Warn:** <N> | **Fail:** <N>
**Rollbacks:** <N>
### Failure Patterns
<Most common failure reasons>
---
## Issues Discovered
### Codegraph Bugs (<count>)
<List with severity, description, context>
### Tooling Issues (<count>)
<List>
### Process Suggestions (<count>)
<List>
### Codebase Observations (<count>)
<List>
---
## Domains Analyzed
<From GLOBAL_ARCH.md — domain map with final status>
---
## Pipeline Freshness
**Main at RECON:** <mainSHA short>
**Main at CLOSE:** <current SHA short>
**Commits behind:** <N>
**Overall staleness:** <fresh|acceptable|stale>
### Drift Events
<Table from drift-report.json: phase, staleness level, impacted targets, action taken>
### Stale Targets
<List of targets whose audit results may not reflect current main — carried forward for next run>
---
## Recommendations for Next Run
<Based on remaining hot spots, unresolved issues, staleness, and patterns observed>
Step 8 — Create PRs (unless --dry-run)
If $ARGUMENTS contains --dry-run:
- Print the PR split plan as a table
- Print the report path
- Stop: "Dry-run complete. Review the plan and report, then re-run
/titan-close to create PRs."
Otherwise, for each PR group (in dependency order):
8a. Create a branch for each PR
git checkout -b titan/<concern>/<domain> main
8b. Cherry-pick the commits
git cherry-pick <sha1> <sha2> ...
If a cherry-pick conflicts:
- Note the conflict in the report
- Skip the problematic commit:
git cherry-pick --skip
- Add the skipped commit to the next PR or flag for manual resolution
8c. Push and create the PR
git push -u origin titan/<concern>/<domain>
gh pr create --title "<title>" --body "$(cat <<'EOF'
## Summary
<bullets from the grouping plan>
## Titan Audit Context
- **Phase:** <concern>
- **Domain:** <domain>
- **Commits:** <count>
- **Depends on:** <PR #N or "none">
## Changes
<file list with brief descriptions>
## Metrics Impact
<relevant before/after metrics for files in this PR>
## Test plan
- [ ] CI passes (lint + build + tests)
- [ ] codegraph check --cycles --boundaries passes
- [ ] No new functions above complexity thresholds
EOF
)"
Record each PR URL.
8d. Return to the original branch
git checkout <original-branch>
Step 9 — Update the report with PR URLs
Edit the report to add the actual PR URLs to the PR split plan table.
Also write a machine-readable summary:
Write .codegraph/titan/close-summary.json:
{
"phase": "close",
"timestamp": "<ISO 8601>",
"version": "<package version>",
"branch": "<branch name>",
"reportPath": "generated/titan/<report-name>.md",
"metrics": {
"baseline": { "qualityScore": 0, "functionsAboveThreshold": 0, "deadSymbols": 0, "cycles": 0 },
"final": { "qualityScore": 0, "functionsAboveThreshold": 0, "deadSymbols": 0, "cycles": 0 }
},
"audit": { "totalAudited": 0, "pass": 0, "warn": 0, "fail": 0, "decompose": 0 },
"gate": { "totalRuns": 0, "pass": 0, "warn": 0, "fail": 0, "rollbacks": 0 },
"grind": { "targetsProcessed": 0, "adopted": 0, "failed": 0, "falsePositives": 0, "deadSymbolDelta": 0 },
"issues": { "codegraph": 0, "tooling": 0, "process": 0, "codebase": 0 },
"prs": [
{ "number": 0, "url": "<url>", "title": "<title>", "concern": "<type>", "domain": "<domain>", "commits": 0 }
],
"commits": { "total": 0, "cherryPickFailures": 0 }
}
Step 10 — Final cleanup
-
Update titan-state.json: set currentPhase to "close", update lastUpdated.
-
Snapshot cleanup: If the pipeline is fully complete (all PRs created):
codegraph snapshot delete titan-baseline 2>/dev/null
Delete any remaining batch snapshots.
-
Branch cleanup: Delete all titan working branches now that their content lives in focused PR branches. These branches accumulate indefinitely without this step and leave hundreds of orphaned commits that never reach main.
List what will be deleted:
git branch --list 'refactor/titan-*' 'docs/titan-*'
Delete locally (force-delete is safe — all work is in the created PR branches). The current worktree branch cannot be deleted while checked out; it will be cleaned up when the worktree is torn down. Skip it gracefully with || true:
git branch --list 'refactor/titan-*' | xargs -r -I{} git branch -D {} || true
git branch --list 'docs/titan-*' | xargs -r -I{} git branch -D {} || true
Delete from remote:
git branch --list 'refactor/titan-*' | sed 's/^[* ]*//' | xargs -r git push origin --delete 2>/dev/null || true
git branch --list 'docs/titan-*' | sed 's/^[* ]*//' | xargs -r git push origin --delete 2>/dev/null || true
Print count of branches deleted.
-
Titan reports are committed to the repo (not gitignored). The generated/titan/ directory is tracked so reports are preserved in git history.
Step 11 — Report to user
Print:
TITAN CLOSE — Pipeline Complete
Report: generated/titan/<report-name>.md
Metrics Delta:
Quality Score: X → Y (+Z)
Functions above threshold: X → Y (-Z)
Dead symbols: X → Y (-Z)
Audit: <N> audited — <P> pass, <W> warn, <F> fail
Gate: <N> runs — <P> pass, <W> warn, <F> fail, <R> rollbacks
Issues: <N> logged (<B> codegraph bugs, <T> tooling, <S> suggestions)
PRs Created: <N>
#1: <title> (<url>)
#2: <title> (<url>)
...
Merge order: PR #1 → #2 → #3 (respect dependencies)
Next: merge PRs in order, then /titan-reset to clean up artifacts.
Rules
- Always use
--json and -T on codegraph commands.
- Never paste raw JSON into your response — parse and extract.
- Dry-run is safe. Only
--dry-run omission triggers branch creation and PR submission.
- Cherry-pick failures are noted, not fatal. Skipped commits get flagged in the report.
- One PR = one concern. Never mix dead code removal with refactors in the same PR.
- Dependency order matters. Create and label PRs so reviewers know the merge sequence.
- The report is the deliverable. It must be comprehensive enough for someone who wasn't present to understand what happened.
- Issue tracker is append-only. Never modify issues from other phases — only compile them.
Self-Improvement
This skill lives at .claude/skills/titan-close/SKILL.md. Adjust PR grouping logic or report structure after dogfooding.