่ทณ่ฝ‰่‡ณ

๐Ÿ–ฅ๏ธ Boring Monitoring Tools

Real-time monitoring of your Agent's brain and body state.

Boring provides three ways to monitor your autonomous loop, depending on your environment and needs.

๐Ÿ› ๏ธ Interface Comparison

Boring interfaces are categorized into TUI (Terminal-based) and Dashboard (Web-based).

Type Name Trigger Command Trigger Type Core Features
TUI Boring Flow boring flow Automatic (via boring start) Shows state-machine activity, Architect/Builder/Healer progress.
TUI Boring Watch boring watch Manual Real-time file monitoring with Automatic Ruff/Lint checks and AI fix suggestions.
TUI Boring Monitor boring-monitor Manual Real-time Token cost, success rates, ASCII dependency graphs.
Component RAG Watcher (Background) Automatic Invisible Feature: Auto-detects file changes and updates RAG vector index silently.
Web Boring Dashboard boring-dashboard Manual Web Analytics: Token trends, Quality Score history, Log Explorer.

๐Ÿš€ Interface Detailed Guide

1. ๐Ÿ“Š Boring Monitor (TUI)

The fastest, most direct way to monitor. No browser needed.

Trigger: 

boring-monitor

4. ๐Ÿ‘๏ธ Boring Watch (TUI)

A dedicated monitoring mode that listens for code changes: - Auto-check: Automatically runs ruff, mypy, or tests whenever you save a file. - AI Fix: Displays immediate AI fix suggestions in the terminal when errors are detected.

boring watch

๐Ÿง  Automatic RAG Indexing (Hidden Feature)

There is also an "invisible trigger". Boring V13+ includes a RAG Watcher. While you are in a boring start loop, the system automatically detects your file changes and updates the local vector database (RAG). This ensures the AI always has the latest code context without requiring manual re-indexing.

2. ๐Ÿ‰ Boring Flow (TUI)

Automatically enabled when running boring start or the "One Dragon" workflow. No manual trigger required during active loops.

Trigger: 

boring flow
# OR automatically via start
boring start

3. ๐Ÿค– Boring Dashboard (Web/Streamlit)

Best for beautiful analytics and deep RAG Brain inspection.

Trigger: 

boring-dashboard
Note: If the command isn't found, try python -m boring dashboard.

โœจ Core Features (Dashboard)

๐Ÿ“Š Live Status

  • Loop Count: Current autonomous loop iteration.
  • Status: Is the Agent THINKING, CODING, or SLEEPING.
  • API Calls: Monitor token usage and API call frequency.
  • Personal Stats: See your top tools and usage patterns. Learn more.

๐Ÿ” Usage Analytics (New in V11.5)

The Monitor now includes a dedicated Usage Stats tab for introspection. - See which tools you use most. - Identify efficiency bottlenecks. - View data powered by boring_usage_stats.

๐Ÿ›‘ Circuit Breaker

  • View breaker status (CLOSED/OPEN).
  • If the Agent gets stuck or fails repeatedly, the breaker trips to protect your API Quota.
  • Anomaly Detection: View loop anomalies blocked by the P5 Safety Net. Learn more.

๐Ÿง  Brain Explorer

See your Agent's RAG Memory and Learning Patterns. - Learned Patterns: See what "Error->Fix" patterns the Agent has learned. - Project Context: View which files and structures are indexed by RAG.

๐Ÿ“ Live Logs

  • Watch the Agent's thought process and debugging logs in real-time.

๐Ÿ› ๏ธ Troubleshooting & FAQ

1. Packages "not installed" despite pip success?

This usually happens when you have multiple Python versions (e.g., 3.12 and 3.13) installed on Windows. Your pip might be installing to a different version than the global boring command uses.

Solution: Run as a Python module Use python -m to force the use of the Python environment where you installed the packages: - Launch Dashboard: python -m boring dashboard / python -m boring.monitor --web - Launch Monitor: python -m boring.monitor

2. "Streamlit is required" Error

If running boring-dashboard prompts that streamlit is missing, ensure you installed the GUI extras: 

pip install "boring-aicoding[gui]"
# Or the full version
pip install "boring-aicoding[all]"
If you are using a virtual environment, ensure it is activated.

3. "No such command 'monitor'"

Remember that monitor is not a subcommand. Use boring-monitor (with a hyphen) instead.

4. Missing "tree-sitter-languages"

This is used for advanced code parsing. If it's not detected after installation, use the python -m boring method described above.

5. Why don't my changes take effect?

This usually happens because the MCP Server hasn't reloaded your code. - Solution: In your IDE (e.g., Cursor) MCP Settings, click the "Refresh" button next to your connection. After modifying any Python tool code, you must restart the server for the AI to learn new skills or see logic updates.

Pro Tip: Keep the monitor open while coding, just like a monitoring screen in a hacker movie! ๐Ÿ˜Ž