feat: update .gitignore to include flows directory and adjust clawteam templates exclusion; enhance LEARNINGS.md and RULES.md with critical spawn method guidelines; revise TOOLS.md to emphasize tmux usage; modify SKILL.md files to prohibit subprocess backend; increase memory limit in docker-compose.yml

.gitignore

@@ -25,6 +25,8 @@
 !.openclaw/canvas/.gitkeep
 .openclaw/update-check.json
 
+.openclaw/flows/
+
 # ==========================================
 # 3. WORKSPACE ISOLATION (GIỮ KHUNG SƯỜN)
 # ==========================================

.openclaw/workspace/.learnings/LEARNINGS.md

@@ -294,3 +294,40 @@ User explicitly required: khi review code thì phải cực kỳ khắt khe, ph
 - First-Seen: 2026-04-06
 - Last-Seen: 2026-04-06
 ---
+
+---
+
+## [LRN-20260406-001] clawteam_spawn_method
+
+**Logged**: 2026-04-06T10:10:00Z
+**Priority**: critical
+**Status**: resolved
+**Area**: orchestration
+
+### Summary
+**Never use subprocess for spawning agents.** Always use `clawteam spawn tmux openclaw --team <team> --agent-name <name> --agent-type <type> --task "<task>"`.
+
+### Details
+When spawning ClawTeam agents, using `clawteam spawn subprocess openclaw` causes agents to die immediately after the subprocess shell exits. This resulted in:
+- All 12 agents spawning but dying instantly
+- 75 tasks all remaining `pending` with 0 `in_progress` because no one was alive to work
+- Tasks falsely marked `completed` (only directory copying happened, not actual development)
+- Zero dependency tracking (all tasks independent despite clear ordering requirements)
+- Complete project management failure
+
+### Suggested Action
+- **ALWAYS** use `clawteam spawn tmux openclaw --team <team> --agent-name <name> ...` for persistent agent sessions
+- **NEVER** use `clawteam spawn subprocess` — it exits immediately after the shell terminates
+- Verify agents are alive: `tmux list-sessions | grep clawteam`
+- Check task progress: `clawteam task list <team> --status in_progress`
+- Set task dependencies with `clawteam task update <team> <id> --add-blocked-by <dep-id>`
+- Monitor progress: `clawteam board attach <team>`
+
+### Metadata
+- Source: error
+- Tags: clawteam, spawn, subprocess, orchestration, critical
+- Pattern-Key: clawteam.spawn_method
+- Recurrence-Count: 1
+- First-Seen: 2026-04-06
+- Last-Seen: 2026-04-06
+

.openclaw/workspace/RULES.md

@@ -149,5 +149,20 @@ When PM receives escalated issue:
 4. If re-assigning: new worker gets full context (error logs, target files, what was tried)
 5. Never add rework to existing blocked task — either unblock it or reassign entirely
 
-### 6. Usage Pattern
+### 6. Spawn Method (ABSOLUTE RULE — ZERO EXCEPTIONS)
+**ALWAYS** use `clawteam spawn tmux openclaw --team <team> --agent-name <name> --agent-type <type> --task "<task>"`
+**NEVER EVER** use `clawteam spawn subprocess` — subprocess agents die immediately after the shell exits, resulting in:
+- All spawned agents dying instantly (0 workers alive)
+- Task counts stuck at 0 in_progress, 75+ tasks never started
+- False "completed" statuses (files created by manual copy, not by agents)
+- Zero dependency tracking despite clear ordering requirements
+- Complete project management failure
+
+**Verification after spawning:**
+- `tmux list-sessions | grep clawteam` — expect 1 session per agent
+- `clawteam task list <team> --status in_progress` — expect tasks actively working
+- `clawteam board attach <team>` — monitor progress in real-time
+- `clawteam task list <team> --status blocked` — check for blocked tasks and resolve
+
+### 7. Usage Pattern
 When the user requests to use ClawTeam, do NOT use `sessions_spawn` with `runtime="acp"`. Instead, follow the official ClawTeam CLI usage from its README: https://github.com/HKUDS/ClawTeam. Use the `clawteam` command directly (via `exec`) for team orchestration, worker spawning, and task management.

.openclaw/workspace/TOOLS.md

@@ -58,12 +58,16 @@ clawteam team spawn-team <team-name> -d "<description>" -n leader
 # Then create tasks matching template.tasks
 clawteam task create <team> "<task>" -o <owner>
 # Spawn workers matching template.agents
-clawteam spawn subprocess openclaw --team <team> --agent-name <name> --task "..."
+# ❌ WRONG: NEVER use subprocess — agents die immediately after shell exits
+# clawteam spawn subprocess openclaw --team <team> --agent-name <name> --task "..."
+
+# ✅ CORRECT: always use tmux openclaw for persistent agent sessions
+clawteam spawn tmux openclaw --team <team> --agent-name <name> --task "..."
 ```
 
 ## ClawTeam Multi-Agent Orchestration
 
-> **Important:** ClawTeam CLI command is `clawteam`. ALWAYS specify `tmux openclaw` or `subprocess openclaw` as backend+agent — default `claude` is not available.
+> **Important:** ClawTeam CLI command is `clawteam`. **ALWAYS** specify `tmux openclaw` as the backend — **NEVER** use `subprocess`. Default `claude` is not available.
 > Official repo: https://github.com/HKUDS/ClawTeam
 
 ### ⚠️ CRITICAL: Spawn Syntax
@@ -246,8 +250,11 @@ clawteam task create <team> "<task>" -o <owner>
 clawteam spawn tmux openclaw --team <team> --agent-name <name> --agent-type coder \
   --task "Do X. After done: clawteam task update <team> <TASK_ID> --status completed"
 
-# Or with subprocess (faster, no tmux monitor)
+# ❌ WRONG — subprocess agents die immediately; never use this
-clawteam spawn subprocess openclaw --team <team> --agent-name <name> --task "..."
+# clawteam spawn subprocess openclaw --team <team> --agent-name <name> --task "..."
+
+# ✅ ALWAYS use tmux (persistent, monitorable, survives network blips)
+# clawteam spawn tmux openclaw --team <team> --agent-name <name> --task "..."
 
 # View board
 clawteam board show <team>
@@ -259,3 +266,11 @@ clawteam task update <team> <task-id> --status completed
 - `--model` (model is set by OpenClaw config, not clawteam)
 - `--timeout` (no timeout flag on spawn)
 - `-m` (conflicts with clawteam's own options)
+
+### 📌 Critical Reminders
+- **ALWAYS use `tmux openclaw`** — `subprocess` agents die immediately
+- **Set task dependencies**: `clawteam task update <team> <id> --add-blocked-by <dep-id>`
+- **Verify agents alive**: `tmux list-sessions | grep clawteam`
+- **Monitor progress**: `clawteam board attach <team>`
+- **Check in-progress tasks**: `clawteam task list <team> --status in_progress`
+- **Unblock stuck tasks**: use `clawteam task update <team> <id> --status pending` after resolving blocker

.openclaw/workspace/skills/clawteam-dev/SKILL.md

@@ -214,9 +214,9 @@ ls ~/.clawteam/workspaces/ 2>&1 | grep -q "No such file" && echo "OK: no workspa
 
 Add `--transport p2p` and `export CLAWTEAM_TRANSPORT=p2p` before Step 3 to test ZeroMQ direct messaging with file fallback. The rest of the steps remain the same.
 
-### With subprocess Backend
+### ⚠️ Do NOT use subprocess backend
 
-Replace `clawteam spawn` with `clawteam spawn subprocess claude` to test the subprocess backend instead of tmux. Note: `board attach` and tmux verification steps won't apply.
+The `subprocess` backend causes agents to die immediately after the shell exits. **Always use the tmux backend** for persistent agent sessions.
 
 ## What This Test Validates
 
@@ -224,7 +224,7 @@ Replace `clawteam spawn` with `clawteam spawn subprocess claude` to test the sub
 |-----------|-----------|
 | `team spawn-team` | Team creation with leader |
 | `task create --blocked-by` | Task dependency chains |
-| `spawn` (tmux backend) | Agent process launch with worktree isolation |
+| `spawn` (tmux backend only) | Agent process launch with worktree isolation |
 | Identity propagation | Env vars passed to sub-agents |
 | Agent coordination | Workers update task status and send messages to leader |
 | Auto-unblock | Blocked task unblocked when dependencies complete |

.openclaw/workspace/skills/clawteam-multi-agent-coordination/SKILL.md

@@ -122,7 +122,7 @@ clawteam board live my-team --interval 3
 
 | Setting | Default | Override |
 |---------|---------|----------|
-| Backend | `tmux` | `clawteam spawn subprocess ...` |
+| Backend | `tmux` only | Use `clawteam spawn tmux openclaw` — subprocess NOT allowed |
 | Command | `claude` (default) / `openclaw` (ours) | `clawteam spawn tmux openclaw` for our setup |
 | Workspace | `auto` (git worktree) | `--no-workspace` or config `workspace=never` |
 | Permissions | skip | `--no-skip-permissions` or config `skip_permissions=false` |

docker-compose.yml

@@ -8,7 +8,7 @@ services:
       resources:
         limits:
           cpus: '4'
-          memory: 8G
+          memory: 10G
         reservations:
           memory: 2G
     environment: