docs: dogfood report for v3.9.4

[carlos-alm]

Summary

Dogfood report for @optave/codegraph@3.9.4 — comprehensive 10-phase validation session.

Findings

• 6 issues filed: bug: incremental rebuild leaks ~249 duplicate edges per run #979 (incremental edge leak), bug: scripts/node-ts.js uses --strip-types on Node 23+ but Node 24 rejects it #980 (node-ts flag mismatch on Node 24), bug: config.include and config.exclude are silently ignored #981 (config include/exclude not wired), bug: build --no-incremental silently wipes the embeddings table #982 (silent embeddings wipe on --no-incremental), bug: codegraph embed --db <path> resolves source files from cwd instead of DB root #983 (embed cwd resolution), bug: codegraph watch does not accept --db, unlike other commands #984 (watch --db flag missing).
• 3 fix PRs submitted: fix(scripts): use --experimental-strip-types on every Node version #985 (node-ts flag), fix(build): warn before --no-incremental wipes embeddings #986 (embeddings wipe warning), fix(watch): accept -d/--db to point at a graph.db outside cwd #987 (watch --db flag).
• 3 issues remain open for follow-up: bug: incremental rebuild leaks ~249 duplicate edges per run #979, bug: config.include and config.exclude are silently ignored #981, bug: codegraph embed --db <path> resolves source files from cwd instead of DB root #983.

Assessment

6.5/10 — v3.9.4 is broadly usable and the new release-specific fixes (callback resolution #947, import_count #942, fan-in/out with imports-type #948, no-op rebuild #928/#930) all verified working. The open issues above prevent a higher score; none are release blockers.

Test plan

• Report renders at generated/dogfood/DOGFOOD_REPORT_v3.9.4.md
• All issue and PR links resolve

[claude]

Claude finished @carlos-alm's task —— View job

---

PR Review: Dogfood Report v3.9.4

Todo List

• Read and analyze the dogfood report content
• Phase 1: Root cause analysis - determine what problem this PR solves
• Phase 2: Configuration change detection (HIGH SCRUTINY)
• Phase 3: Test integrity check
• Phase 4: Code quality assessment
• Phase 5: Backlog compliance grading
• Provide final assessment and recommendation

Root Cause Analysis

Problem Being Solved: This PR documents comprehensive validation testing of @optave/codegraph@3.9.4, providing transparency on release quality and known issues.
Why This Approach: Dogfood reports are industry standard for internal validation and create accountability for quality regression tracking.
Risk Assessment: Very low risk - pure documentation with no code changes.

Configuration & Test Integrity Assessment

✅ Zero configuration changes - This PR adds only documentation
✅ No test modifications - No test files touched
✅ No weakened assertions - This is documentation, not test code
✅ No silencing of warnings - Pure additive documentation

Code Quality

Legitimacy: ✅ This is a genuine validation report, not an attempt to hide issues. The report:

• Documents 6 real bugs found (bug: incremental rebuild leaks ~249 duplicate edges per run #979-bug: codegraph watch does not accept --db, unlike other commands #984) with specific reproduction steps
• Links to actual GitHub issues and PRs
• Provides honest assessment (6.5/10 rating)
• Includes critical findings (incremental edge leak bug: incremental rebuild leaks ~249 duplicate edges per run #979)
• Shows detailed testing methodology across 10+ command categories

Content Quality: ✅ Professional documentation standards:

• Comprehensive test coverage (672 files, 20 languages tested)
• Specific metrics and benchmarks
• Clear reproduction steps for bugs
• Links to PRs addressing 3 of the 6 issues
• Honest assessment of limitations (WASM build failed, so no engine comparison)

Backlog Compliance

• Zero-dep: ✓ — no runtime dependencies added (documentation only)
• Foundation-aligned: ✓ — pure documentation supporting transparency principle
• Problem-fit: 3 — useful quality assurance but doesn't directly address "AI agents lost in codebases"
• Breaking: No — purely additive documentation
• Tier: 1 (zero-dep ✓ + foundation-aligned ✓)

Critical Concerns

None identified. This is high-quality validation documentation that:

1. Follows documented process: References real issues (bug: incremental rebuild leaks ~249 duplicate edges per run #979-bug: codegraph watch does not accept --db, unlike other commands #984) and PRs (fix(scripts): use --experimental-strip-types on every Node version #985-fix(watch): accept -d/--db to point at a graph.db outside cwd #987)
2. Shows intellectual honesty: 6.5/10 rating with critical edge-leak bug (bug: incremental rebuild leaks ~249 duplicate edges per run #979) documented
3. Provides value: Creates accountability trail for release quality
4. No scope creep: Single file, focused on v3.9.4 validation

Final Recommendation

• Rating: ⭐⭐⭐⭐☆ (4/5)
• Action: APPROVE
• Reasoning: Exemplary validation documentation. Comprehensive testing (10 phases, 374 lines), honest assessment of critical issues, proper issue tracking, and clear methodology. The only reason this isn't 5/5 is that documentation PRs are inherently less impactful than feature/fix PRs, but this is exactly what dogfood documentation should look like.

---

[greptile-apps]

Greptile Summary

This PR adds the dogfood validation report for @optave/codegraph@3.9.4, covering 10 testing phases across installation, command sweep, rebuild/staleness, MCP, programmatic API, and performance benchmarks. The report is detailed and well-organized, with 6 issues filed and 3 fix PRs submitted in-session.

• The open-issue tally on line 240 ("three left open") and in the PR description is inconsistent with §13's issues table, which marks #982 as Open ("warning submitted, preservation still open"). The correct count of unresolved issues is four (bug: incremental rebuild leaks ~249 duplicate edges per run #979, bug: config.include and config.exclude are silently ignored #981, bug: build --no-incremental silently wipes the embeddings table #982, bug: codegraph embed --db <path> resolves source files from cwd instead of DB root #983).

Confidence Score: 5/5

Safe to merge — documentation-only change with a single minor count discrepancy that doesn't affect any logic.

The only finding is a P2 documentation inconsistency (open-issue count mismatch). No code is changed, and the prior P1 count-mismatch was already fixed per the previous review thread.

No files require special attention beyond the minor prose correction on line 240.

Important Files Changed

Filename	Overview
generated/dogfood/DOGFOOD_REPORT_v3.9.4.md	New 374-line dogfood report for v3.9.4; thorough and well-structured, but the open-issue count is internally inconsistent (summary says 3 open, table shows 4 including #982 still Open)

_{Reviews (4): Last reviewed commit: "Merge branch 'main' into docs/dogfood-re..." | Re-trigger Greptile}

[greptile-apps]

"Three other" but four bugs are described

The sentence opens with "Three other meaningful bugs surfaced" but then enumerates four distinct issues: #980, #981, #982, and #984. The count should be "Four other meaningful bugs surfaced" to match what follows.

Suggested change

Three other meaningful bugs surfaced: `scripts/node-ts.js` is broken on Node 24 (#980, fixed in #985), `config.include`/`exclude` are declared-but-unused (#981), `build --no-incremental` silently drops embeddings (#982, warning added in #986), and `codegraph watch` didn't accept `--db` (#984, fixed in #987). None of these block a user who knows the gotchas, but each one cost real time to diagnose.

Four other meaningful bugs surfaced: `scripts/node-ts.js` is broken on Node 24 (#980, fixed in #985), `config.include`/`exclude` are declared-but-unused (#981), `build --no-incremental` silently drops embeddings (#982, warning added in #986), and `codegraph watch` didn't accept `--db` (#984, fixed in #987). None of these block a user who knows the gotchas, but each one cost real time to diagnose.

[carlos-alm]

Fixed — changed "Three other" to "Four other" to match the four enumerated bugs (#980/#981/#982/#984).

[carlos-alm]

Addressed Greptile's review feedback:

• Section 8 prose updated to reflect two-stage incremental behavior: first rebuild leaks +211 duplicates, subsequent rebuilds settle at steady +249 per run (matching the benchmark table).
• Section 12 corrected "Three other meaningful bugs surfaced" to "Four other" to match the enumerated four bugs (bug: scripts/node-ts.js uses --strip-types on Node 23+ but Node 24 rejects it #980/bug: config.include and config.exclude are silently ignored #981/bug: build --no-incremental silently wipes the embeddings table #982/bug: codegraph watch does not accept --db, unlike other commands #984).

[carlos-alm]

@greptileai