# Memory index for agent-restore-harness

Key facts to carry across sessions (memory API was unavailable during session 2026-04-22):

## What was done (2026-04-22)

- Repo created at `/home/m2/agent-restore-harness`, committed to git (main)
- Nasr (Smithers) workspace path typo FIXED: openclaw.json now points to `/home/developer/.openclaw/workspace`
- Nasr workspace files snapshotted to `agents/nasr/` (SOUL, USER, IDENTITY, MEMORY, AGENTS, HEARTBEAT, skills.txt)
- Forgejo push PENDING — needs API token created at git.machinemachine.ai (m2 user, id=1)

## What's on spark3

```
/home/m2spark3/telegram/
  m2.zip             64MB → 85MB uncompressed, 38 HTML files (FULL m2 DM history)
  parlobyg.zip       41MB → Parlo full chat
  machine.machine.zip  572KB → MM group chat (6 HTML files, GMI clinic context)
  m2-devops.zip      79KB
  agent.memory.system/ → ingest/telegram_export.py ready to use
```

## spark4 status

- 22,744 points in agent_memory collection (Qdrant ok)
- Redis degraded — RDB write failure — READ ONLY

## Nasr skills that need Forgejo repos

clinical-pathway, patient-pathway, quantum-trading, trading-app-dev,
2dexy-sync, ml-training, xcode-remote, build-verify, cursor-agent

Fleet-wide skills (likely already in machine-machine org):
m2-memory, rlm-memory, unified-search, agency-agents, harness-engine,
intent-elicit, intent-router, spec-discovery, mm-pdf

## Restore order

nasr (in progress) → parlo → peter → m2