docs: add Titan audit report for v3.15.0 run

[carlos-alm]

Summary

Adds the final Titan Paradigm audit report for the v3.15.0 self-audit run (RECON → GAUNTLET → SYNC → FORGE → GRIND → PARITY → CLOSE), documenting before/after metrics, the 30-phase FORGE execution (including a mid-run session-crash recovery), GRIND's dead-helper adoption results, and the 10-PR split of this run's 39 commits.

Titan Audit Context

• Phase: close (documentation)
• Domain: generated/titan
• Commits: 1
• Depends on: none — informational only, does not touch product code

Changes

• generated/titan/titan-report-v3.15.0-2026-07-04T08-17-55.md (new)

Test plan

• CI passes (lint + build + tests) — docs-only change, no functional impact

[greptile-apps]

Greptile Summary

Adds the final Titan Paradigm audit report (generated/titan/titan-report-v3.15.0-2026-07-04T08-17-55.md) for the v3.15.0 self-audit run, covering all seven pipeline phases (RECON → CLOSE), a mid-run session-crash recovery, 30-phase FORGE execution, GRIND dead-helper adoption, and the 10-PR split of 39 commits. This is a docs-only addition with no impact on product code.

• The report documents a net -45 reduction in above-threshold functions, a correctly characterised dead-symbol increase (+497, attributed to known classifier false positives filed as issues codegraph fails to attribute call edges to CodeGraph class getters/methods, causing false dead-unresolved flags #1769–codegraph exports does not credit instanceof ClassName checks as consumers #1784), and a clean PARITY pass with zero main-branch drift over the ~50-hour run.
• One internal inconsistency exists: the GRIND Results header states "False positives: 10" (matching the 10-row table) while the Metrics note derives 15 from the same 24-candidate pool (24 − 9 adopted = 15), leaving the summary fields unbalanced.

Confidence Score: 5/5

Docs-only addition of a single markdown report file — no product code, configuration, or build artifacts touched.

The change adds one documentation file with no functional impact. The single finding is a numeric inconsistency within the report itself between two different sections counting false positives, which does not affect the repo's code, tests, or CI.

No files require special attention — only generated/titan/titan-report-v3.15.0-2026-07-04T08-17-55.md is changed, and it is purely informational.

Important Files Changed

Filename	Overview
generated/titan/titan-report-v3.15.0-2026-07-04T08-17-55.md	New 310-line Titan audit report for v3.15.0; metrics arithmetic is mostly consistent (39-commit PR table checks out, GATE pass/warn/fail sums to 39, issues totals add up), but the GRIND Results header reports 10 false positives while the Metrics note derives 15 from the same 24-candidate pool (24 − 9 adopted = 15), leaving 5 targets unaccounted in the header.

Flowchart

%%{init: {'theme': 'neutral'}}%%
flowchart TD
    RECON["RECON\n1h07m\n886 files, 18168 symbols"] --> GAUNTLET
    GAUNTLET["GAUNTLET\n2h03m\n46 targets audited"] --> SYNC
    SYNC["SYNC\n12m\n30-phase plan"] --> FORGE
    FORGE["FORGE\n~14h48m active\n30/30 phases, 29 commits\n⚡ crash @ phase 28→29"] --> GRIND
    GRIND["GRIND\n9h46m\n30/30 phases\n6 adoption commits\n9 adopted/promoted\n10 FP batches logged"] --> PARITY
    PARITY["PARITY\n15m\nClean — 0 drift events"] --> GATE
    GATE["GATE\n39 inline runs\n24 pass / 11 warn / 4 fail\n1 rollback"] --> CLOSE
    CLOSE["CLOSE\n24m\n10 PRs created\n6 issues filed\nThis report"] --> DONE["39 commits → 10 PRs\n#1785–#1794"]

Loading

%%{init: {'theme': 'base', 'themeVariables': {"darkMode": true, "background": "#0d1117", "primaryColor": "#21262d", "primaryTextColor": "#e6edf3", "primaryBorderColor": "#8b949e", "lineColor": "#8b949e", "textColor": "#e6edf3", "edgeLabelBackground": "#161b22", "actorBkg": "#21262d", "actorBorder": "#8b949e", "actorTextColor": "#e6edf3", "actorLineColor": "#8b949e", "signalColor": "#8b949e", "signalTextColor": "#e6edf3", "noteBkgColor": "#373320", "noteBorderColor": "#d4a72c", "noteTextColor": "#f0e6c0", "labelBoxBkgColor": "#21262d", "labelBoxBorderColor": "#8b949e", "labelTextColor": "#e6edf3", "loopTextColor": "#e6edf3", "activationBkgColor": "#30363d", "activationBorderColor": "#8b949e"}}}%%
flowchart TD
    RECON["RECON\n1h07m\n886 files, 18168 symbols"] --> GAUNTLET
    GAUNTLET["GAUNTLET\n2h03m\n46 targets audited"] --> SYNC
    SYNC["SYNC\n12m\n30-phase plan"] --> FORGE
    FORGE["FORGE\n~14h48m active\n30/30 phases, 29 commits\n⚡ crash @ phase 28→29"] --> GRIND
    GRIND["GRIND\n9h46m\n30/30 phases\n6 adoption commits\n9 adopted/promoted\n10 FP batches logged"] --> PARITY
    PARITY["PARITY\n15m\nClean — 0 drift events"] --> GATE
    GATE["GATE\n39 inline runs\n24 pass / 11 warn / 4 fail\n1 rollback"] --> CLOSE
    CLOSE["CLOSE\n24m\n10 PRs created\n6 issues filed\nThis report"] --> DONE["39 commits → 10 PRs\n#1785–#1794"]

Loading

_{Reviews (3): Last reviewed commit: "fix: correct issue-count arithmetic in T..." | Re-trigger Greptile}

[greptile-apps]

The inline list contains 18 distinct issue numbers (#1720–#1776), but the surrounding sentence says "17 already filed." Together with the 6 newly filed at CLOSE, that would make the stated total of 23 off by one (should be 24). The parenthetical acknowledges cross-referenced entries, but each number in the list is a distinct GitHub issue — if one is intentionally double-counted rather than a unique issue, it should be called out explicitly so the total remains internally consistent.

Suggested change

	23 GitHub issues total are attributable to this run's findings: **17 already filed during RECON/GAUNTLET/GRIND** (#1720, #1721, #1723, #1724, #1725, #1726, #1727, #1728, #1729, #1730, #1731, #1741, #1760, #1761, #1769, #1771, #1773, #1776 — several are cross-referenced by multiple issue-tracker entries), and **6 newly filed at CLOSE** after a duplicate check against the open issue list:
	24 GitHub issues total are attributable to this run's findings: **18 already filed during RECON/GAUNTLET/GRIND** (#1720, #1721, #1723, #1724, #1725, #1726, #1727, #1728, #1729, #1730, #1731, #1741, #1760, #1761, #1769, #1771, #1773, #1776 — several are cross-referenced by multiple issue-tracker entries), and **6 newly filed at CLOSE** after a duplicate check against the open issue list:

Note: If this suggestion doesn't match your team's coding style, reply to this and let me know. I'll remember it for next time!

[carlos-alm]

Fixed — corrected "17" to "18" already-filed issues (matching the 18 numbers enumerated) and updated the total from 23 to 24.

[carlos-alm]

@greptileai