π The original agent harness. Deploy intelligent multi-player swarms, coordinate autonomous workflows, and build conversational AI systems. Features adaptive memory, self-learning intelligence, federation, vector RAG integration, and native Claude Code / Codex / Hermes and many more Integrated
V3 Statusline & Daemon Integration
Overview
The V3 hooks system integrates with background daemons and statusline displays to provide real-time monitoring of swarm activity, implementation progress, and security status. This document covers the integration between hooks and the daemon infrastructure.
Architecture
βββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββ
β Claude Code Session β
βββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββ€
β SessionStart Hook β
β ββ> .claude/helpers/daemon-manager.sh start β
β ββ> swarm-monitor.sh (process detection, 3s interval) β
β ββ> metrics-db.mjs (SQLite sync, 30s interval) β
β ββ> hooks-daemon.mjs (learning sync, 60s interval) β
βββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββ€
β Hooks System Integration β
β ββ> PreToolUse Hook β Activity detection β Metrics update β
β ββ> PostToolUse Hook β Outcome recording β Pattern learning β
β ββ> PreCommand Hook β Risk assessment β Security tracking β
β ββ> PostTask Hook β Progress update β Statusline refresh β
βββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββ€
β statusline.sh (on-demand, <200ms) β
β ββ> Reads from: β
β ββ .claude-flow/metrics.db (SQLite, primary) β
β ββ .claude-flow/hooks.db (ReasoningBank patterns) β
β ββ .claude-flow/metrics/*.json (exported, legacy compat) β
βββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββ€
β SessionEnd Hook β
β ββ> .claude/helpers/daemon-manager.sh stop β
β ββ> Final metrics export, pattern consolidation β
βββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββ
Daemons
1. Daemon Manager (daemon-manager.sh)
Central control for all background processes with hooks integration.
# Start all daemons (called by SessionStart hook)
.claude/helpers/daemon-manager.sh start [swarm_interval] [metrics_interval] [hooks_interval]
# Default intervals: swarm=3s, metrics=30s, hooks=60s
.claude/helpers/daemon-manager.sh start 3 30 60
# Stop all daemons (called by SessionEnd hook)
.claude/helpers/daemon-manager.sh stop
# Restart with new configuration
.claude/helpers/daemon-manager.sh restart
# Check daemon status
.claude/helpers/daemon-manager.sh status
PID Management:
.claude-flow/pids/
βββ swarm-monitor.pid # Swarm monitoring daemon
βββ metrics-daemon.pid # Metrics collection daemon
βββ hooks-daemon.pid # Hooks learning daemon
Log Files:
.claude-flow/logs/
βββ daemon.log # Daemon manager operations
βββ swarm-monitor.log # Process detection logs
βββ metrics-daemon.log # Metrics sync logs
βββ hooks-daemon.log # Learning operation logs
2. Metrics Database Daemon (metrics-db.mjs)
SQLite-based metrics storage with hooks integration.
# Sync metrics from implementation
node .claude/helpers/metrics-db.mjs sync
# Export to JSON (statusline compatibility)
node .claude/helpers/metrics-db.mjs export
# Run as daemon
node .claude/helpers/metrics-db.mjs daemon [interval_seconds]
# Query specific metrics
node .claude/helpers/metrics-db.mjs query "SELECT * FROM hooks_metrics"
Extended Schema for Hooks:
-- Hooks Activity Metrics
CREATE TABLE hooks_metrics (
id INTEGER PRIMARY KEY,
total_executions INTEGER,
total_failures INTEGER,
avg_execution_time REAL,
patterns_learned INTEGER,
routing_confidence REAL,
last_updated TEXT
);
-- Per-Hook Statistics
CREATE TABLE hook_stats (
hook_name TEXT PRIMARY KEY,
category TEXT,
execution_count INTEGER,
success_rate REAL,
avg_time_ms REAL,
last_executed TEXT
);
-- Routing History
CREATE TABLE routing_history (
id INTEGER PRIMARY KEY,
task_hash TEXT,
recommended_agent TEXT,
confidence REAL,
was_successful INTEGER,
timestamp TEXT
);
-- Learning Patterns
CREATE TABLE learning_patterns (
pattern_id TEXT PRIMARY KEY,
category TEXT,
quality_score REAL,
usage_count INTEGER,
created_at TEXT,
last_used TEXT
);
3. Swarm Monitor Daemon (swarm-monitor.sh)
Real-time process detection with hooks event emission.
# Single check (returns JSON)
.claude/helpers/swarm-monitor.sh check
# Continuous monitoring with hooks notification
.claude/helpers/swarm-monitor.sh monitor [interval]
# Status with hooks metrics
.claude/helpers/swarm-monitor.sh status --include-hooks
Output:
{
"timestamp": "2025-01-05T12:00:00Z",
"swarm": {
"agentic_flow_processes": 3,
"mcp_server_processes": 2,
"estimated_agents": 5,
"active": true,
"coordination_active": true
},
"hooks": {
"daemon_running": true,
"patterns_loaded": 156,
"routing_enabled": true
}
}
4. Hooks Learning Daemon (hooks-daemon.mjs)
Background process for ReasoningBank consolidation.
# Run consolidation cycle
node .claude/helpers/hooks-daemon.mjs consolidate
# Run as daemon
node .claude/helpers/hooks-daemon.mjs daemon [interval_seconds]
# Export patterns
node .claude/helpers/hooks-daemon.mjs export --format json
Operations:
- Trajectory pruning: Remove low-quality trajectories
- Pattern consolidation: Merge similar patterns
- Memory optimization: Compress old patterns
- Index rebuilding: Optimize HNSW index
Statusline Integration
Format
β Claude Flow V3 β agentic-flow@alpha β β v3
βββββββββββββββββββββββββββββββββββββββββββββββββββββ
ποΈ DDD Domains [βββββ] 5/5 β‘ 1.0x β 2.49x-7.47x
π€ Swarm Agents β [ 5/15] π’ CVE 3/3 πΎ 156 patterns
π§ Architecture DDD β93% β Security βCLEAN β Hooks βACTIVE
π Routing 89% accuracy β Avg 4.2ms β 1547 operations
βββββββββββββββββββββββββββββββββββββββββββββββββββββ
Statusline Script
#!/bin/bash
# .claude/statusline.sh - V3 with Hooks Integration
# Read from SQLite if available
if [ -f ".claude-flow/metrics.db" ]; then
METRICS=$(node -e "
const db = require('.claude/helpers/metrics-db.mjs');
console.log(JSON.stringify(db.getStatuslineData()));
" 2>/dev/null)
# Extract values
DDD_PROGRESS=$(echo "$METRICS" | jq -r '.ddd_progress // 0')
ACTIVE_AGENTS=$(echo "$METRICS" | jq -r '.active_agents // 0')
CVE_STATUS=$(echo "$METRICS" | jq -r '.cve_status // "PENDING"')
PATTERNS=$(echo "$METRICS" | jq -r '.patterns_learned // 0')
ROUTING_ACC=$(echo "$METRICS" | jq -r '.routing_accuracy // 0')
HOOK_STATUS=$(echo "$METRICS" | jq -r '.hooks_status // "INACTIVE"')
fi
# Format output
printf "β Claude Flow V3 β agentic-flow@alpha β β v3\n"
printf "βββββββββββββββββββββββββββββββββββββββββββββββββββββ\n"
printf "ποΈ DDD Domains %s β β‘ Performance targets active\n" "$DDD_PROGRESS"
printf "π€ Swarm Agents β [%2d/15] π’ CVE %s πΎ %d patterns\n" "$ACTIVE_AGENTS" "$CVE_STATUS" "$PATTERNS"
printf "π§ Hooks %s β Routing %d%% β Learning %s\n" "$HOOK_STATUS" "$ROUTING_ACC" "ACTIVE"
printf "βββββββββββββββββββββββββββββββββββββββββββββββββββββ\n"
Performance
| Component | Target | Typical |
|---|---|---|
| statusline.sh total | <200ms | 138ms |
| SQLite query | <50ms | 12ms |
| JSON export | <100ms | 45ms |
| Hooks metrics | <20ms | 8ms |
Hook Configuration
Claude Settings (~/.claude/settings.json)
{
"hooks": {
"SessionStart": [
{
"hooks": [
{
"type": "command",
"timeout": 5000,
"command": "/workspaces/claude-flow/.claude/helpers/daemon-manager.sh start 3 30 60"
}
]
}
],
"SessionEnd": [
{
"hooks": [
{
"type": "command",
"timeout": 3000,
"command": "/workspaces/claude-flow/.claude/helpers/daemon-manager.sh stop"
}
]
}
],
"PreToolUse": [
{
"hooks": [
{
"type": "command",
"timeout": 100,
"command": "node .claude/helpers/hooks-daemon.mjs notify-activity"
}
]
}
]
},
"statusLine": {
"type": "command",
"command": "/workspaces/claude-flow/.claude/statusline.sh"
}
}
V3 Project Settings (.claude-flow/config.json)
{
"daemons": {
"enabled": true,
"swarmMonitor": {
"enabled": true,
"interval": 3000
},
"metricsSync": {
"enabled": true,
"interval": 30000,
"database": ".claude-flow/metrics.db"
},
"hooksLearning": {
"enabled": true,
"interval": 60000,
"database": ".claude-flow/hooks.db"
}
},
"statusline": {
"enabled": true,
"refreshOnHook": true,
"showHooksMetrics": true,
"showSwarmActivity": true
}
}
Event Flow
1. Session Start
SessionStart Hook Triggered
β
ββ> daemon-manager.sh start
β ββ> Start swarm-monitor.sh (every 3s)
β ββ> Start metrics-db.mjs daemon (every 30s)
β ββ> Start hooks-daemon.mjs (every 60s)
β
ββ> Initialize ReasoningBank
β ββ> Load patterns from .claude-flow/hooks.db
β ββ> Warm HNSW index for retrieval
β
ββ> First statusline render
2. During Session
Tool Use (e.g., Edit)
β
ββ> PreToolUse Hook
β ββ> hooks-daemon.mjs notify-activity
β ββ> Increment metrics counter
β
ββ> Tool Execution
β
ββ> PostToolUse Hook
ββ> Record outcome to ReasoningBank
ββ> Update hooks_metrics table
ββ> (If successful) Queue pattern distillation
3. Background Sync
Every 30 seconds (metrics-db.mjs daemon):
β
ββ> Scan V3 implementation files
β ββ> Count files per module
β ββ> Calculate DDD progress
β
ββ> Check security status
β ββ> CVE remediation count
β
ββ> Export JSON for statusline
β
ββ> Log sync completion
Every 60 seconds (hooks-daemon.mjs):
β
ββ> Consolidate ReasoningBank patterns
β ββ> Prune low-quality trajectories
β ββ> Merge duplicate patterns
β
ββ> Update hook_stats table
β
ββ> Optimize HNSW index if needed
4. Session End
SessionEnd Hook Triggered
β
ββ> daemon-manager.sh stop
β ββ> SIGTERM to all daemons
β ββ> Wait for graceful shutdown
β
ββ> Final ReasoningBank consolidation
β ββ> Export learned patterns
β
ββ> Final metrics sync
β ββ> Export to JSON
β
ββ> Log session summary
Metrics Tracked
Hooks Performance
| Metric | Description | Update Frequency |
|---|---|---|
total_executions |
Total hook invocations | Real-time |
total_failures |
Failed hook executions | Real-time |
avg_execution_time |
Average hook time (ms) | Every 30s |
patterns_learned |
ReasoningBank patterns | Every 60s |
routing_accuracy |
Task routing success rate | Every 60s |
routing_confidence |
Average routing confidence | Every 60s |
Per-Hook Statistics
| Metric | Description |
|---|---|
hook_name |
Hook identifier |
category |
pre-edit, post-edit, routing, etc. |
execution_count |
Times executed |
success_rate |
Success percentage |
avg_time_ms |
Average execution time |
last_executed |
Timestamp of last execution |
Learning Metrics
| Metric | Description |
|---|---|
trajectories_stored |
Total trajectories in ReasoningBank |
patterns_distilled |
Memories extracted from trajectories |
consolidation_runs |
Pattern consolidation cycles |
index_size |
HNSW index entry count |
avg_retrieval_time |
Pattern retrieval latency |
Troubleshooting
Daemons Not Starting
# Check daemon status
.claude/helpers/daemon-manager.sh status
# View logs
tail -f .claude-flow/logs/daemon.log
# Check for stale PID files
ls -la .claude-flow/pids/
# Manual cleanup and restart
rm .claude-flow/pids/*.pid
.claude/helpers/daemon-manager.sh start
Statusline Not Updating
# Force metrics sync
node .claude/helpers/metrics-db.mjs sync
# Check SQLite database
sqlite3 .claude-flow/metrics.db "SELECT * FROM hooks_metrics"
# Verify statusline script
bash -x .claude/statusline.sh
ReasoningBank Issues
# Check hooks database
node .claude/helpers/hooks-daemon.mjs status
# Force consolidation
node .claude/helpers/hooks-daemon.mjs consolidate --force
# Rebuild patterns
node .claude/helpers/hooks-daemon.mjs rebuild-index
High CPU from Daemons
# Increase daemon intervals
.claude/helpers/daemon-manager.sh restart 10 60 120
# Disable non-essential daemons
.claude/helpers/daemon-manager.sh start --no-hooks-daemon
Files Reference
.claude/
βββ statusline.sh # Main statusline script
βββ settings.json # Claude settings with hooks
βββ helpers/
βββ daemon-manager.sh # Daemon lifecycle
βββ metrics-db.mjs # Metrics SQLite engine
βββ swarm-monitor.sh # Process detection
βββ hooks-daemon.mjs # Learning background process
.claude-flow/
βββ metrics.db # Main metrics database
βββ hooks.db # ReasoningBank storage
βββ config.json # V3 configuration
βββ pids/
β βββ swarm-monitor.pid
β βββ metrics-daemon.pid
β βββ hooks-daemon.pid
βββ logs/
β βββ daemon.log
β βββ swarm-monitor.log
β βββ metrics-daemon.log
β βββ hooks-daemon.log
βββ metrics/
βββ v3-progress.json # Exported V3 metrics
βββ swarm-activity.json # Exported swarm status
βββ hooks-metrics.json # Exported hooks metrics