Phase 2: Accuracy & Analysis Depth — cast timeline, healer metrics, game-data regen, regression net

.planning/PROJECT.md

@@ -33,6 +33,11 @@ A player pastes a Warcraft Logs URL and instantly gets **accurate**, actionable
 - ✓ GDPR consent layer: Google Privacy & Messaging (TCF v2.2) gates PostHog capture and session replay for EEA/UK visitors; non-EEA unaffected (MONY-01) — Phase 1
 - ✓ `/privacy` and `/terms` pages (individual operator, California law, contact info@lootlistplus.com) — Phase 1 (quick task 260906-kzw)
 - ✓ OPS-01 ship gate as a repeatable document (`docs/OPS-01-SHIP-GATE.md`) with `seo-invariants` local-vs-prod diff; Phase 1 gate closed with GSC + PostHog evidence — Phase 1
+- ✓ Per-fight cast Timeline tab: ordered cast log with idle-gap rows, death marker, per-ability filter chips, windowed rendering and truncation notice, behind a paginated/rate-limited/cached `/api/timeline` (ACC-03) — Phase 2
+- ✓ Healer-relevant analysis: effective HPS / overheal % / uptime from one shared helper on both the player card and the raid Healer Breakdown, plus healer-only suggestion rules thresholded against top healers (ACC-04) — Phase 2
+- ✓ Game data as generated truth: `npm run regen-game-data` wago.tools pipeline, three era modules with Classic/TBC-first collision precedence, `lib/cla-constants.ts` a thin re-export, `docs/GAME-DATA-AUDIT.md` as the standing review artifact (ACC-01) — Phase 2
+- ✓ Regression net: recorded WCL fixtures drive tests for `cla-engine`, `raid-overview-engine`, `wcl-client`, timeline and healer engines (14 files / 135 tests); a red suite now blocks a deploy in the OPS-01 gate (ACC-02) — Phase 2
+- ✓ Phase 2 shipped to production (`dpl_5bwk1fJJNuZXkoC5poPGFZQpGy6c`) after a preview sweep; PR #15 open for review — Phase 2
 
 ### Active
 
@@ -55,7 +60,8 @@ A player pastes a Warcraft Logs URL and instantly gets **accurate**, actionable
 - [ ] Ship first monetization — ads as default plan; evaluate low-effort alternatives before committing
 
 **Product quality:**
-- [ ] Improve analysis accuracy (top priority per user) and expand tool capabilities
+- [ ] Expand tool capabilities beyond the Phase 2 accuracy work (per-player share image bakes in the healer percentile basis — a `costly` decision to revisit in Phase 3)
+- [ ] Close the Phase 2 code-review findings: CR-01 healer suggestions fire when `hasHealing` is false (wrong recommendation, live), WR-01 timeline "player not in fight" check is report-scoped not fight-scoped, WR-02 StrictMode double-fires `timeline_filter_used`, WR-03 stale precedence comment in `game-data.test.ts` (`.planning/phases/02-accuracy-analysis-depth/02-REVIEW.md`)
 
 **Ecosystem & community:**
 - [ ] Connect LootList+ ↔ ParseForge (cross-promotion between the two sites)
@@ -83,8 +89,8 @@ A player pastes a Warcraft Logs URL and instantly gets **accurate**, actionable
   - Site-wide CTR is structurally misleading (navigational impressions dominate) — never treat it as a health metric; segment first
   - Title template appends " | ParseForge" (13 chars) — child page titles must stay under ~47 chars; always render pages to verify, don't eyeball metadata
 - **Next keyword candidate**: `wow log analyzer` (392 impr/28d, pos 3.5, tool intent) — segment before building
-- **Known concerns** (from CONCERNS.md): CSP still report-only; core engines (cla, raid-overview, wcl-client) largely untested. *(Resolved Phase 1: PostHog EU consent flow now gated by the Google CMP.)* Deferred from Phase 1: formal ASVS security review — gsd security tooling not installed; per-plan STRIDE registers exist
-- **Manual follow-ups after Phase 1**: paste `https://parseforge.gg/privacy` into AdSense → Privacy & messaging → message site settings; re-check PostHog event definitions (`theme_changed`, `consent_resolved`) at the Phase 2 gate; `git push origin main` (unpushed phase forces sequential execution)
+- **Known concerns** (from CONCERNS.md): CSP still report-only. *(Resolved Phase 1: PostHog EU consent flow now gated by the Google CMP. Resolved Phase 2: core engines now under a fixture-driven regression net.)* Deferred from Phases 1 and 2: formal ASVS security review — gsd security tooling not installed in this profile; per-plan STRIDE registers exist
+- **Manual follow-ups after Phase 2**: paste `https://parseforge.gg/privacy` into AdSense → Privacy & messaging → message site settings (still open); PostHog event definitions (`timeline_viewed`, `timeline_error`, `timeline_filter_used`, `analysis_complete`, `theme_changed`, `consent_resolved`) — confirmed by the developer at the Phase 2 UAT, but the PostHog MCP connection returns `INVALID_API_KEY` and its active project is LootList+ App, so re-wire it before the Phase 3 gate; GSC shows `/` and `/analyze/*` as "Crawled – currently not indexed" with pre-deploy crawl dates — request recrawl of the Phase 2 build and watch `/` specifically; `origin/main` is at `55d2010` (pre-Phase-2) and Phase 2 lives on PR #15 (`growth/phase-2-accuracy-depth`) — merge or push to restore parallel worktree execution; two game-data ids (96264, 96294) flagged for human review in `.planning/WINDOWS.md`
 - **Feedback source today**: owner's personal Discord (uncomfortable) — motivates the community requirement
 
 ## Constraints
@@ -112,6 +118,11 @@ A player pastes a Warcraft Logs URL and instantly gets **accurate**, actionable
 | Class-attribute theming via next-themes, tokens enforced by `theme-parity` + `token-audit` scripts | Machine-checkable DSGN-01 so the Phase 7 redesign can't regress hardcoded colours | ✓ Good (Phase 1) |
 | Preview deploy before every prod deploy when no real-browser pass happened | Agents can't drive a browser; a preview lets the developer eyeball before visitors do | ✓ Good (Phase 1) |
 | Defer the formal ASVS security review for Phase 1 | Security tooling absent from this gsd-core profile; STRIDE registers exist per plan | — Pending (install tooling, run `/gsd-secure-phase 01`) |
+| Record real WCL responses as committed fixtures before building on assumed shapes (Phase 2 Wave 0) | RESEARCH.md carried five unverified API-shape assumptions; recording once fed both the timeline and the regression net, and corrected A3 (scoped healing table has per-ability `overheal` but no `activeTime`) | ✓ Good (Phase 2) |
+| Era precedence for generated game data is Classic/TBC-first, not later-era-wins | Client builds reuse numeric IDs across eras (806 enchant + 295 gem collisions); later-wins silently corrupted pinned pairs 3003/2667/32196 | ✓ Good (Phase 2) — 130/178 consumable names are explicit sourced overrides |
+| One healer-metrics helper feeds both the player card and the raid overview | D-08: two surfaces must never disagree; parity is asserted by test and was confirmed live in production | ✓ Good (Phase 2) |
+| Preview-first deploy plus push + PR before shipping Phase 2 | Regenerated game-data names had the least automated coverage; PR #15 makes the regen diff reviewable (D-12) | ✓ Good (Phase 2) — prod `dpl_5bwk1fJJNuZXkoC5poPGFZQpGy6c` |
+| Defer the formal ASVS security review for Phase 2 (same as Phase 1) | `gsd-secure-phase` still not installed in this profile; per-plan STRIDE registers exist | — Pending (install tooling, run `/gsd-secure-phase 02`) |
 
 ## Evolution
 
@@ -131,4 +142,4 @@ This document evolves at phase transitions and milestone boundaries.
 4. Update Context with current state
 
 ---
-*Last updated: 2026-09-07 after Phase 1*
+*Last updated: 2026-09-08 after Phase 2*

.planning/REQUIREMENTS.md

@@ -37,10 +37,10 @@ Requirements for this milestone. Each maps to roadmap phases.
 
 ### Accuracy & Capabilities
 
-- [ ] **ACC-01**: Game-data accuracy re-audit via the wago.tools regeneration workflow (no hand-typed ID maps)
-- [ ] **ACC-02**: Test coverage added for the untested engines (`cla-engine`, `raid-overview-engine`, `wcl-client`)
-- [ ] **ACC-03**: User can view a per-fight cast timeline (closes table-stakes gap vs competitors)
-- [ ] **ACC-04**: Healer-specific analysis improved (healer metrics beyond raw HPS)
+- [x] **ACC-01**: Game-data accuracy re-audit via the wago.tools regeneration workflow (no hand-typed ID maps)
+- [x] **ACC-02**: Test coverage added for the untested engines (`cla-engine`, `raid-overview-engine`, `wcl-client`)
+- [x] **ACC-03**: User can view a per-fight cast timeline (closes table-stakes gap vs competitors)
+- [x] **ACC-04**: Healer-specific analysis improved (healer metrics beyond raw HPS)
 
 ### Community & Cross-Promotion
 
@@ -103,10 +103,10 @@ Which phases cover which requirements. Updated during roadmap creation.
 | MONY-01 | Phase 1 | Complete |
 | MONY-02 | Phase 4 | Pending |
 | MONY-03 | Phase 4 | Pending |
-| ACC-01 | Phase 2 | Pending |
-| ACC-02 | Phase 2 | Pending |
-| ACC-03 | Phase 2 | Pending |
-| ACC-04 | Phase 2 | Pending |
+| ACC-01 | Phase 2 | Complete |
+| ACC-02 | Phase 2 | Complete |
+| ACC-03 | Phase 2 | Complete |
+| ACC-04 | Phase 2 | Complete |
 | COMM-01 | Phase 5 | Pending |
 | COMM-02 | Phase 5 | Pending |
 | OPS-01 | Phase 1 | Phase 1 gate closed 2026-09-06 — docs/OPS-01-SHIP-GATE.md (standing gate, re-run every phase) |

.planning/ROADMAP.md

@@ -20,7 +20,7 @@ ParseForge is a live product with real organic traffic, so this milestone ships
 Decimal phases appear between their surrounding integers in numeric order.
 
 - [x] **Phase 1: Foundation — Themes & Consent** - Token audit unlocks a dark/light toggle; consent layer gates ads and PostHog replay (completed 2026-09-06)
-- [ ] **Phase 2: Accuracy & Analysis Depth** - Verified game data, tested engines, cast timeline, healer metrics
+- [x] **Phase 2: Accuracy & Analysis Depth** - Verified game data, tested engines, cast timeline, healer metrics (completed 2026-09-08)
 - [ ] **Phase 3: Share Loop** - Roast/award cards, per-player permalinks, share CTA that survives later phases
 - [ ] **Phase 4: Ads Live** - AdSense in reserved, consent-gated slots that never crowd the core flow
 - [ ] **Phase 5: Community & Cross-Promotion** - Dedicated Discord, LootList+ two-way links, auto-posted reports
@@ -92,33 +92,33 @@ Plans:
   4. `cla-engine`, `raid-overview-engine`, and `wcl-client` have automated tests that fail when analysis output changes — a regression safety net in place before the redesign touches anything.
   5. New timeline and healer surfaces ship with PostHog events and pass the GSC verification pass (OPS-01 gate).
 
-**Plans**: 9 plans
+**Plans**: 9/9 plans executed
 **UI hint**: yes
 
 Plans:
 **Wave 1**
 
-- [ ] 02-01-PLAN.md — Wave 0: record real WCL fixtures, then the query, type and rate-limit contracts the phase builds on
+- [x] 02-01-PLAN.md — Wave 0: record real WCL fixtures, then the query, type and rate-limit contracts the phase builds on
 
 **Wave 2** *(blocked on Wave 1 completion)*
 
-- [ ] 02-02-PLAN.md — Cast timeline tracer: open the Timeline tab and read your casts in order, end to end
-- [ ] 02-03-PLAN.md — wago.tools regeneration pipeline: script, overrides layer, three generated era modules
+- [x] 02-02-PLAN.md — Cast timeline tracer: open the Timeline tab and read your casts in order, end to end
+- [x] 02-03-PLAN.md — wago.tools regeneration pipeline: script, overrides layer, three generated era modules
 
 **Wave 3** *(blocked on Wave 2 completion)*
 
-- [ ] 02-04-PLAN.md — Healer metrics: effective HPS, overheal and uptime from one shared helper on both surfaces
-- [ ] 02-05-PLAN.md — Timeline depth: idle gaps, death marker, filter chips, virtualisation, truncation notice
-- [ ] 02-06-PLAN.md — Cut cla-constants over to generated data and emit docs/GAME-DATA-AUDIT.md
+- [x] 02-04-PLAN.md — Healer metrics: effective HPS, overheal and uptime from one shared helper on both surfaces
+- [x] 02-05-PLAN.md — Timeline depth: idle gaps, death marker, filter chips, virtualisation, truncation notice
+- [x] 02-06-PLAN.md — Cut cla-constants over to generated data and emit docs/GAME-DATA-AUDIT.md
 
 **Wave 4** *(blocked on Wave 3 completion)*
 
-- [ ] 02-07-PLAN.md — Healer-specific suggestions relative to top-healer values; DPS-shaped rules gated off
-- [ ] 02-08-PLAN.md — Engine regression tests for cla-engine, raid-overview-engine and wcl-client
+- [x] 02-07-PLAN.md — Healer-specific suggestions relative to top-healer values; DPS-shaped rules gated off
+- [x] 02-08-PLAN.md — Engine regression tests for cla-engine, raid-overview-engine and wcl-client
 
 **Wave 5** *(blocked on Wave 4 completion)*
 
-- [ ] 02-09-PLAN.md — OPS-01 gate close-out and approval-gated production deploy
+- [x] 02-09-PLAN.md — OPS-01 gate close-out and approval-gated production deploy
 
 **Notes**: Accuracy is the project's stated core value ("a wrong recommendation is worse than no recommendation"), so it precedes monetization and redesign. The engine tests here are what make Phase 7's redesign safe to attempt.
 
@@ -224,7 +224,7 @@ Phases execute in numeric order: 1 → 2 → 3 → 4 → 5 → 6 → 7
 | Phase | Plans Complete | Status | Completed |
 |-------|----------------|--------|-----------|
 | 1. Foundation — Themes & Consent | 9/9 | Complete    | 2026-09-06 |
-| 2. Accuracy & Analysis Depth | 0/9 | Planned      | - |
+| 2. Accuracy & Analysis Depth | 9/9 | Complete    | 2026-09-08 |
 | 3. Share Loop | 0/TBD | Not started | - |
 | 4. Ads Live | 0/TBD | Not started | - |
 | 5. Community & Cross-Promotion | 0/TBD | Not started | - |