clawd/memory/working-context.md

77 lines
3.1 KiB
Markdown

# Working Context
*Updated: 2026-03-15 21:00 ET (nightly maintenance)*
## Status: Sunday evening — channel migration day, Matter pairing, voice-in exploration
---
## Current Projects
### inou.com — Active Development
- **Committed state:** Multiple auto-commits pushed earlier today
- **inou:** large backlog of files, 2 unpushed commits historically — check `git status` before working
- **james-dashboard:** Synced and pushed to Zurich (12 commits earlier today including data/state)
- **clawd:** Pushed to Zurich with Discord-only channel config + cron/disk checks in HEARTBEAT
### Channel Migration — COMPLETE (2026-03-15)
- **Discord → PRIMARY** as of today. Telegram disabled in openclaw.json.
- Signal was already retired 2026-03-01
- Telegram retired 2026-03-15
- TUI remains available as fallback (useful in tmux alongside Claude Code)
- **MEMORY.md updated** to reflect Discord as primary
### George (vault1984 Writer Agent)
- Live on Discord. No active development today.
---
## Today's Events (2026-03-15 Sunday)
### Matter Pairing (Home Assistant)
- Johan was re-pairing Matter smart plugs/lights
- **Bedroom 1:** Successfully re-paired after HAOS reboot
- Entities: `light.bedroom_1_light_1/2/3`, power sensors
- Johan confirmed "finally added" around 19:04 ET
- **Bedroom P (Primary/Master Bedroom):** Was showing `unavailable` before EOD
- Entities expected: `light.bedroom_p_light_1/2/3`, current/energy/power/voltage sensors
- Status at end of session: in progress (Johan was going to pair)
- ⚠️ **Pending** — confirm Bedroom P Matter device is paired and automations restored
### Channel Config Applied
- Removed Telegram + Signal from openclaw.json
- Restarted gateway (had timeout, but confirmed healthy afterward)
- Verified: `telegram: enabled=False`, `signal: enabled=False`, `discord: enabled=True`
### Voice Input Discussion
- Johan asked about voice interaction with James via Discord
- Found: @TomRChen's OpenClaw skill (published same day) — Discord voice DM → faster-whisper → James
- **Voice-in only** (no TTS back). Johan already has Wispr Flow so voice→text is covered.
- **Gap identified:** TTS output (James speaking back) — not yet built. FishAudio mentioned as option.
- No implementation started — Johan just exploring.
### daily-updates.service Fix
- Was failing due to LLM version compare logic
- Fixed earlier in day — service shows `Result=success`
### Git Commits Pushed
- james-dashboard: 12+ commits synced to Zurich
- clawd: feat commit with Discord-only config, cron/disk checks in HEARTBEAT, CLAUDE.md updates
---
## Open Threads / Pending Items
1. **Bedroom P Matter pairing** — confirm completed and automations intact
2. **Voice TTS for James** — explore FishAudio or other TTS for Discord voice output (Johan interested but not urgently)
3. **inou feature backlog** — side-by-side DICOM comparison view is next priority
4. **Claude Code** — updated to 2.1.76 tonight
---
## Key Infrastructure State
- **Gateway:** healthy, Discord-only, port 18789
- **MC (message-bridge):** healthy
- **Disk:** 54% used
- **HAOS:** rebooted today for Matter pairing, back online
- **OCR service:** running on forge