xamination is mandatory after conclusions.** Not optional. Every conclusion recorded triggers a `reflect()` pass. ### What the Persona Cannot Do - Cannot mark significance as "consequence" without linking to a real decision entity - Cannot mark confidence as "high" without verified evidence - Cannot skip self-examination - Cannot delete or modify its own prior reflections (append-only applies to self-exams too) --- ## SAFEGUARD 7: DEPLOYMENT READINESS (CONFIRMED) Environment already supports everything Phase 1 needs: | Requirement | Status | Action Needed | |-------------|--------|---------------| | PostgreSQL server | Active (v18.6) | None | | Database | `epistles` exists | Create tables only | | Node pg driver | Installed (^8.13.1) | None | | Application role | Not yet created | Create `solidivism_app` role | | Schema changes | None yet | Run DDL (new tables only) | | Server config changes | None needed | None | | New dependencies | None needed | None | **Pre-implementation checklist (no changes yet):** - [x] PostgreSQL active and reachable - [x] Target database exists - [x] Node driver available - [x] No new packages needed - [ ] Create restricted role (on authorization) - [ ] Create tables (on authorization) - [ ] Verify role permissions (on authorization) --- ## EXACT PHASE 1 CHANGE BOUNDARIES ### Files Modified | File | Change | Lines | Reversible | |------|--------|-------|------------| | `foundation.js` | Add `observe()` function + 3 call sites | ~15 lines added, 0 modified | Yes: delete added lines | | `foundation.js` | Add `epistemology: Epistemology` to return | 1 line added | Yes | | `foundation.js` | Append Epistemology closure (new code at end) | ~400 lines added, 0 existing modified | Yes: delete appended block | | `index.html` | Add `Epistemology.init()` after `Solidivism.init()` | 1 line added | Yes | | Database | Create 3 tables + 1 role + triggers | DDL only, no existing tables touched | Yes: DROP TABLE (with backup) | ### Files NOT Modified - No changes to rendering pipeline - No changes to avatar construction - No changes to input handling - No changes to camera math - No changes to collision logic - No changes to existing `addBuilding()` behavior (only added optional param) - No changes to any other epistles.xyz routes - No changes to server.js (new API routes added, none modified) ### Lines of Existing Code Touched: ZERO Phase 1 is purely additive to existing code. The only "modifications" are: - 3 `observe()` call insertions (new lines, not changed lines) - 1 return object addition (new property) - 1 init call (new line) Every existing line of 3D code remains byte-identical. --- ## REMAINING DEPLOYMENT RISKS ### Risk 1: localStorage Quota Exceeded **Likelihood:** Low | **Impact:** Medium If user is offline for extended session, event cache may exceed 5MB localStorage limit. **Mitigation:** Circular buffer (keep most recent 5,000 events). Older unsynced events marked as "potentially lost" in next sync. PostgreSQL remains authoritative for what it received. ### Risk 2: Clock Skew Between Browser and Server **Likelihood:** Medium | **Impact:** Low `event_time` (client) may differ from `recorded_at` (server) by seconds or minutes. **Mitigation:** Both timestamps stored. Queries default to `recorded_at` for ordering. `event_time` preserved for client-context. ### Risk 3: API Endpoint Abuse **Likelihood:** Low | **Impact:** Medium Public POST endpoint could be spammed with fake events. **Mitigation (Phase 1):** Rate limiting (100 events/min per IP). Phase 2: authentication. **Note:** Fake events are marked with `source: 'browser'` and can be filtered. They cannot modify existing records (append-only). ### Risk 4: Persona Generates Excessive Questions **Likelihood:** Medium | **Impact:** Low Unconstrained curiosity could generate thousands of question entities. **Mitigation:** Question creation throttled (max 10/min). Questions require explicit trigger (not automatic for every observation). ### Risk 5: Backfill Data Quality **Likelihood:** High | **Impact:** Low Backfilled entities from investigation transfer will have many "medium" confidence levels due to compaction. **Mitigation:** All backfilled records marked `source: 'backfill'`, `confidence` honestly assessed. They are starting points, not ground truth. ### Risk 6: Performance Impact of observe() **Likelihood:** Low | **Impact:** Medium If `observe()` is called too frequently (e.g., every frame), it could impact FPS. **Mitigation:** Phase 1 has only 3 call sites, none in the per-frame loop. Collision observation throttled to 1/sec. Measured in T9. --- ## AUTHORIZATION REQUIRED The following actions require The Programmer's explicit authorization: 1. Create `solidivism_app` database role 2. Create `solidivism_events`, `solidivism_entities`, `solidivism_edges` tables 3. Deploy modified `foundation.js` to staging (`/solidivism-staging/`) 4. Run backfill script for known entiti