chore(graphify): refresh knowledge graphs

[allyblockcast]

Automated knowledge-graph refresh from the graphify-reindex CronJob.

AST-only rebuild (graphify update), zero LLM tokens. Only the structural graph artifacts changed; no source files were touched. Safe to merge once CI is green.

[allyblockcast]

Ally — Consolidated PR Review

Lenses: pr-review-toolkit (code, tests, comments, errors, types) + gstack/review + native-codex.
Reviewed head: 9e4e9a1

Looks good. This is a pure generated-artifact refresh — 3 files, all under server/src/graphify-out/, no source touched — and every claim the PR body and report make checks out against the artifact itself.

Rather than eyeball a 13.5 MB diff, I validated the artifacts directly at this head:

• Counts are self-consistent. graph.json parses as valid JSON with 9,971 nodes and 20,671 links — exactly the 9971 nodes · 20671 edges claimed in GRAPH_REPORT.md:8.
• Freshness claim is accurate. built_at_commit is fb85860161027986682dcab9abe6c002855b116c, which is the current master tip, matching `fb858601` in GRAPH_REPORT.md:14.
• Referential integrity is clean. 0 links referencing a non-existent node id, 0 duplicate node ids, 0 nodes with a null source_file. The 2 self-loops are the expected recursion artifacts.
• Labels and communities agree exactly. Node community ids span 0–476 with 477 distinct values, matching the 477 entries in .graphify_labels.json — no orphaned labels, no unlabeled communities. The labels diff is purely additive (461–476).
• Report arithmetic holds. 437 shown + 40 thin = 477 (prior: 383 + 38 + 40 = 461).
• No secret or path leakage. Node records are metadata-only (id, label, norm_label, source_file, source_location, community, file_type, _origin) — no code bodies. No credential-shaped strings, and no absolute, /home/, or /Users/ paths.
• No build impact. Nothing in the repo imports graphify-out, and server/tsconfig.json compiles only TypeScript under src, so the 10.4 MB JSON is not pulled into dist.

CI shows no failures at this head (most jobs still pending; policy, staleness, Helm chart, security-review green).

Critical Issues (0)

Important Issues (0)

Suggestions (3)

• [native-codex] server/src/graphify-out/GRAPH_REPORT.md:8 — The report format changed, not just the data: the structural-only omitted category is gone entirely (it accounted for 40 communities in the 2026-08-10 report), and the phrase is dropped from both the Summary line and the ## Communities heading. That is generator behavior drift, not corpus growth. Because the artifact carries no graphify version stamp, a tool upgrade is indistinguishable from a data shift when reviewing these PRs.
  • Consider emitting the graphify version into GRAPH_REPORT.md alongside the build commit, so format changes are attributable in the diff.
• [gstack/review] server/src/graphify-out/graph.json:1 — There is no .gitattributes entry covering server/src/graphify-out/. Marking graph.json as linguist-generated=true (optionally -diff) would collapse the ~100k-line diff in the PR UI and keep the artifact out of repo language stats, making future refreshes reviewable at a glance.
• [pr-review-toolkit] server/src/graphify-out/GRAPH_REPORT.md:8 — Three 2-file cycles dropped out of this run: external-objects.ts ↔ github-external-object-provider.ts, plugin-lifecycle.ts ↔ plugin-loader.ts, and issues.ts ↔ task-watchdogs.ts. That is a good signal if those were genuinely refactored, but worth a sanity check that it is real improvement rather than a side effect of the same generator drift noted above.

Strengths

• Deterministic AST-only rebuild with zero LLM tokens, so the diff is reproducible and carries no model-output trust boundary.
• The artifact is internally consistent and consistent with both its sidecar label file and its human-readable report — the three files cannot silently drift apart.
• Built from the exact master tip, so the graph is not stale on arrival.
• Scope is tight and honest: the PR body claims "no source files were touched," and the file list confirms it.

Recommended Action

1. No Critical issues to fix before merge.
2. No Important issues this cycle.
3. Consider the Suggestions opportunistically — the version stamp (Suggestion 1) is the one with lasting review value.

[allyblockcast]

Ally — Consolidated PR Review

Lenses: pr-review-toolkit (code, tests, comments, errors, types) + gstack/review + native-codex.
Reviewed head: f845a1f

Looks good. This is a generated AST-only artifact refresh limited to server/src/graphify-out/.

Critical Issues (0)

Important Issues (0)

Suggestions (2)

• [gstack/review] server/src/graphify-out/graph.json:1 — Consider marking this generated artifact in .gitattributes with linguist-generated=true (and optionally -diff) so future refreshes remain reviewable and do not affect language statistics.
• [native-codex] server/src/graphify-out/GRAPH_REPORT.md:8 — Consider recording the graphify tool/version alongside the build commit so future report-format or generator changes are distinguishable from corpus changes.

Strengths

• graph.json parses successfully and reports 10,148 nodes and 21,037 links, matching the report.
• Referential integrity is clean: no duplicate node IDs, missing link endpoints, or nodes without source files.
• The report and sidecar labels cover the same 484 community IDs, and the build commit matches the current master tip.
• Scope is tight: only generated graph artifacts changed; no source files were modified.

Recommended Action

1. No Critical issues to fix before merge.
2. No Important issues this cycle.
3. Consider Suggestions opportunistically.