# Cursor Optimization Run Log

**Last Updated:** 2026-04-03

Append-only operational log for manual optimization runs.

## Entry Template

### OPT-RUN-YYYY-MM-DD-NN

- Date (UTC):
- Slice: `A|B`
- Owner:
- State: `planned|running|blocked|done|rolled-back`
- Commands executed:
- KPI snapshot:
  - `exact_glob_overlap_count`:
  - `blog_json_overlap_count`:
  - `pointer_rules_count`:
  - `docs_hot_path_count`:
- Completed tasks (with evidence):
- What changed this run (summary):
- Deferred tasks (why):
- Risk signals observed:
- Rollback action (if any):
- Backlog IDs created:
- Decision IDs referenced:
- Next action:

---

### OPT-RUN-2026-04-03-01

- Date (UTC): 2026-04-03
- Slice: `A`
- Owner: Codex
- State: `done`
- Commands executed: `make cursor-integrity`, `make cursor-benchmark`, `make cursorignore-audit`, golden probes
- KPI snapshot:
  - `exact_glob_overlap_count`: 48
  - `blog_json_overlap_count`: 7
  - `pointer_rules_count`: tracked in benchmark
  - `docs_hot_path_count`: tracked in benchmark
- Completed tasks (with evidence): baseline stabilization and overlap reductions recorded in scorecard/wave summary
- Deferred tasks (why): none
- Risk signals observed: none
- Rollback action (if any): none
- Backlog IDs created: none
- Decision IDs referenced: DEC-0001..DEC-0007 defaults
- Next action: alternate to Slice B in next manual run

---

### OPT-RUN-2026-04-03-02

- Date (UTC): 2026-04-03
- Slice: `B`
- Owner: Codex
- State: `done`
- Commands executed: `make cursor-integrity`, `make cursor-benchmark`, `make cursorignore-audit`, `make cursor-check-overlap`, `make cursor-check-freshness`, docs inventory/redundancy pipeline
- KPI snapshot:
  - `exact_glob_overlap_count`: 48
  - `blog_json_overlap_count`: 7
  - `pointer_rules_count`: 32
  - `docs_hot_path_count`: 5
- Completed tasks (with evidence): continuous program docs/ledgers/configs added; benchmark/threshold/freshness/report scripts implemented; CI staged gate logic added; CODEOWNERS added; all checks passing
- Deferred tasks (why): CI stage variable promotion to stage 2+ deferred to maintain stabilization window
- Risk signals observed: none
- Rollback action (if any): none
- Backlog IDs created: none
- Decision IDs referenced: DEC-0001..DEC-0007 defaults
- Next action: run next manual cycle with alternating Slice A focus and evaluate stage promotion criteria

---

### OPT-RUN-2026-04-03-03

- Date (UTC): 2026-04-03
- Slice: `A`
- Owner: Codex
- State: `done`
- Commands executed: `make cursor-integrity`, `make cursor-benchmark`, `make cursor-check-overlap`, `make cursor-check-freshness`, `make cursor-check-rule-size`, `make cursor-check-skill-quality`, golden probes via `scripts/find-rules-for-file.py`
- KPI snapshot:
  - `exact_glob_overlap_count`: 48
  - `blog_json_overlap_count`: 7
  - `pointer_rules_count`: 35
  - `docs_hot_path_count`: 5
- Completed tasks (with evidence): added `rule-skill-census.py` + generated census outputs; added warning-first rule-size/no-glob and skill-trigger quality checks; normalized 7 skills with scope/fallback sections; pointerized long no-glob rules with archive backlinks; fixed `find-rules-for-file.py` glob matcher for `**`.
- What changed this run (summary): quality-hardening wave completed with non-increasing overlap and improved no-glob/pointer KPIs.
- Deferred tasks (why): CI stage promotion deferred until stability window criteria are met.
- Risk signals observed: none
- Rollback action (if any): none
- Backlog IDs created: none
- Decision IDs referenced: DEC-0008, DEC-0009, DEC-0010, DEC-0011
- Next action: execute next Slice B run and prioritize remaining rule-size/no-glob warnings for safe reductions

---

### OPT-RUN-2026-04-03-04

- Date (UTC): 2026-04-03
- Slice: `A`
- Owner: Codex
- State: `done`
- Commands executed: `make cursor-integrity`, `make cursor-check-overlap`, `make cursor-benchmark`, targeted probes via `scripts/find-rules-for-file.py`
- KPI snapshot:
  - `exact_glob_overlap_count`: 46
  - `blog_json_overlap_count`: 6
  - `pointer_rules_count`: 35
  - `docs_hot_path_count`: 5
- Completed tasks (with evidence): fixed stale/deleted `relatedRules` entries in `faq-cross-surface.mdc`; removed broad blog JSON glob from `faq-cross-surface.mdc`; narrowed ShiftOps frontend rules from broad `lead-capture*.php` and duplicate JS globs to specific `lead-capture-popup.php` + `shiftops*.js`; corrected malformed `relatedDocs` list formatting in both ShiftOps frontend rules.
- What changed this run (summary): reduced unnecessary rule over-triggering while preserving domain routing and passing integrity checks.
- Deferred tasks (why): broader `v2/pages/**/*.php` FAQ routing kept for now to avoid cross-surface regressions; evaluate in later wave with additional probes.
- Risk signals observed: none
- Rollback action (if any): none
- Backlog IDs created: none
- Decision IDs referenced: DEC-0009, DEC-0011
- Next action: continue targeted glob narrowing on safe hotspots from `rule-footprint.py` with per-change golden probes

---

### OPT-RUN-2026-04-03-05

- Date (UTC): 2026-04-03
- Slice: `A`
- Owner: Codex
- State: `done`
- Commands executed: `make cursor-check-rule-size`, `make cursor-integrity`, `make cursor-benchmark`
- KPI snapshot:
  - `exact_glob_overlap_count`: 46
  - `blog_json_overlap_count`: 6
  - `pointer_rules_count`: 38
  - `docs_hot_path_count`: 5
- Completed tasks (with evidence): pointerized `mcp-usage.mdc`, `og-images.mdc`, `partner-resources-assets.mdc`; added corresponding archive companion docs; promoted CI workflow default fallback stage to 2; tightened overlap thresholds to current baseline.
- What changed this run (summary): completed two pending next steps (stage promotion and warning triage) without overlap regression.
- Deferred tasks (why): controlled deletion batch deferred; no verified zero-reference pointerized candidates available.
- Risk signals observed: none
- Rollback action (if any): none
- Backlog IDs created: none
- Decision IDs referenced: DEC-0012, DEC-0013, DEC-0014
- Next action: build evidence-backed zero-reference deletion candidate list before any deletion batch