Skip to main content

Debugging Agents

Debugging Agents

Troubleshooting AI agents in the LLM platform.

Agent Not Starting

Check Agent Status

# List agents
buildkit agents list

# Check specific agent
buildkit agents status tdd-enforcer-001

# View logs
buildkit agents logs tdd-enforcer-001

# Kubernetes
kubectl logs -n agents <agent-pod> -f

Agent Failures

Task Execution Errors

Check execution logs:

# View task logs
buildkit agents task-logs <task-id>

# View Phoenix traces
open http://localhost:6006
# Filter by agent_id

Common issues:

  1. LLM API errors - Check API keys and quotas
  2. Timeout - Increase task timeout
  3. Missing dependencies - Check agent manifest

Performance Issues

Slow Agent Response

# Check ACE score
agent-tracer ace score --agent-id tdd-enforcer-001

# Analyze with ATLAS
agent-tracer atlas analyze --agent-id tdd-enforcer-001

# View traces in Phoenix
open http://localhost:6006

Agent Communication

OSSA Protocol Issues

# Check OSSA compliance
buildkit ossa validate .agents/<agent-name>/manifest.json

# Test agent discovery
buildkit ossa discover --namespace agents

# Check network status
buildkit mesh status

Debugging Tools

Enable Debug Logging

# Set log level
export AGENT_LOG_LEVEL=debug

# Restart agent
buildkit agents restart tdd-enforcer-001

Attach Debugger

// Add breakpoint
debugger;

// Run with inspector
node --inspect src/agent.ts

Related Pages