Skip to content
- META org package created: IDENTITY, SOUL, USER, TEAM, AGENTS, HEARTBEAT, TELEGRAM_GROUPS
- Memory structure initialised:
daily/, decisions/, projects/, ops/
- Project files created: FRANKEL.md, R6RECORDS.md, S1DEFENCE.md
- Ticket board created:
ops/TICKETS.md
- HQ topic mapping finalized: 1/2/4/6/8/10/12/14/16/18 validated in live routing.
- HQ access controls hardened:
groupPolicy=allowlist, groupAllowFrom=[8267352500], group allowFrom=[8267352500], dmPolicy=pairing, configWrites=false.
- Mention gating finalized: topic 1 allow-no-mention; all other HQ topics mention-gated.
- Guardrails added: non-DevOps topics instructed to refuse restart/config edits and route to DevOps with
CONFIRM: <reason>.
- Ops automation installed:
openclaw-hq-daily.timer (09:05), openclaw-hq-weekly.timer (Mon 09:15), persistent user timers enabled.
- Recovery tooling installed:
openclaw_snapshot.sh + openclaw_restore.sh with retention pruning (14 daily / 8 weekly).
- Scaffolded Next.js 15 + Tailwind 4 + TypeScript project
- SQLite DB (better-sqlite3) with tasks + activity schema
- 6 API routes with Zod validation + activity logging
- 3-panel kanban UI (Sidebar | Kanban | Activity Panel)
- @hello-pangea/dnd drag-and-drop between 5 columns
- Seed data: 11 tasks across all status columns
- Build passes: ✅
- Set global heartbeat interval to
30m (agents.defaults.heartbeat.every).
- Set main-agent heartbeat model override to
minimax-portal/MiniMax-M2.5.
- Verified night sprint cron remains active at
0 3 * * * in Europe/London.
- Tuned
agents.defaults.subagents.maxConcurrent from 8 to 3 to reduce stuck/contended runs.
- Added managed skill
subagent-spawn-sop.
- Added mandatory Sub-agent Spawn SOP section to main + PM AGENTS files for cold-start/session-reset consistency.
- Rebound Telegram HQ Research topic 16 from
research to main.
- Updated PM subagent allowlists to include
research so PMs can spawn Scout as a worker.
- Updated AGENTS routing rules: worker templates (
dev-backend, dev-frontend, research) must use sessions_spawn; persistent agents use sessions_send.
- Created 3 coding agent skills:
claude-code-orchestration, codex-cli, gemini-code
- Codex CLI v0.104.0 authenticated via ChatGPT Plus OAuth
- Config:
~/.codex/config.toml with multi_agent=true, high reasoning, workspace-write sandbox
- Gemini CLI v0.29.5 confirmed broken in WSL2 headless (hangs, zero output)
- Updated pm-s1defence MEMORY.md, DAILY_GOALS.md, HEARTBEAT.md for 2026-02-23 solicitor meeting
- Verified topic 6 binding to pm-s1defence agent
- INCIDENT: Invalid
agent key added to topic config crashed gateway. Fixed by removing key.
- Updated
~/.config/systemd/user/mission-control.service:
WorkingDirectory=/home/winnie/projects/mission-controlExecStart=/usr/bin/node /home/winnie/projects/mission-control/node_modules/.bin/next start -p 3001
- Reloaded user daemon and restarted service (
mission-control.service active).
- Restored Tailscale serve mapping:
tailscale serve --bg --https=8080 http://127.0.0.1:3001
- Result: Mission Control back online at
https://desktop-8bb3hd4-1.tailb7bd99.ts.net:8080/.
- HEARTBEAT.md: rewrote heartbeat run protocol to lightweight router pattern
- All 9 agent AGENTS.md: added “Mission Control — Task Updates” section with curl commands
- Added MC sync step to HEARTBEAT.md
- Created
memory/ops/inter-agent-comms.md — full protocol doc
- Fixed context pruning:
reserveTokensFloor 24000 → 40000 in openclaw.json
dev-ops AGENTS.md: 117 → 71 linesdev-backend + dev-frontend: Replaced verbose MC update block with compact 3-line version- Added mandatory
Checkpoint Output section to dev-backend and dev-frontend
- Created
/home/winnie/backups/backup.sh — backs up Mission Control DB + all 8 OpenClaw agent memory SQLites
- Daily cron at 04:30 (after night sprint finishes)
- 7-day local retention, SQLite
.backup API for WAL-safe copies
- Tested: MC backup 44K/47 tasks ✓, agent memories 99MB ✓
- Issue:
gemini -p "prompt" hangs indefinitely in WSL2 (no TTY)
- Root cause: Ink-based terminal UI (React renderer) blocks waiting for terminal dimensions/input
- Status: Upstream bug. Cannot use Gemini CLI in WSL2 headless on v0.29.5.
- Resolution: v0.30.0 (2026-03-01) fixes the WSL2 headless hang.