Phase 2 baseline: full wiki clone with silo audit manifest

Audit summary: 2890(73), PSB(86), Hacker(40), shared(71), agent(10)
Pre-migration snapshot from openclaw-wiki tag: pre-migration

memory/connections-log.md

@@ -0,0 +1,20 @@
+## 2026-05-04 14:31 UTC
+Connections found: 1
+- [HHS-Hackers] "[[M5 + Thunderbolt scaling]]" ↔ [Research] "[[MLX Distributed Inference]]" — Chris asked crash-bot about GPU scaling options, was directed to MLX. Key finding: MLX supports true distributed inference where model layers split across multiple Apple Silicon machines, both working on every token simultaneously. This is the "uber geek play" for scaling — not just two separate machines but a unified inference cluster via Thunderbolt. Relevant to Chris's long-term AI rig planning.
+
+## 2026-05-03 07:51 UTC (Anomaly)
+**ANOMALY: OpenClaw Settings Silent Revert**
+- Filed by psb-gemma 2026-04-21 (open task: memory/tasks/systemic-settings-loss.md)
+- WAL mode silently disables — "settings appear to be applied but silently fail or revert"
+- Discord Message Content Intent also silently fails (hits 4014 DisallowedIntents)
+- Affects all PSB agents
+- **Implication for Professor:** When configuring OpenClaw, verify settings are actually applied, don't trust silent config.=== Connection Scan 2026-05-03 23:08 UTC ===
+Scanning vault for new cross-silo connections...
+329bfd2 Index Team 2890 Mothman robot code for training reference
+65db595 Add historical game manuals: 2024 CRESCENDO, 2025 REEFSCAPE
+c5e3d0a Add FIRST official technical resources as canonical training reference
+b1f45bb Note: read-only access sufficient for statbotics
+418c501 Add Statbotics as FRC analytics with EPA predictions
+New files today: 2024-crescendo.md, 2025-reefscape.md, first-technical-resources.md, 2890-codebase-index.md, mothman-robot-code.md
+Connections found: 1 (TILT data ↔ fermentation monitoring ↔ crash-bot heartbeat)
+---