feat: log tool_use calls in agent-runner for debugging

[neocode24]

Summary

• Log each tool invocation (name + truncated input) from assistant messages in the agent-runner
• Helps diagnose which tools the agent is calling and with what arguments during container execution
• Input is truncated to 200 chars to avoid log bloat

Changes

• container/agent-runner/src/index.ts — Add tool_use logging in the assistant message handler

Example output

Tool use: Bash input={"command":"ls -la /workspace/group","description":"List files"}
Tool use: Read input={"file_path":"/workspace/group/CLAUDE.md"}
Tool use: mcp__nanoclaw__send_message input={"message":"Hello from the agent"}

Test plan

• Verified logging output in container logs during agent execution
• Confirmed truncation works for large tool inputs

🤖 Generated with Claude Code