rajan-cforge

rajan-cforge / ai-runtime-monitor

Public

CrowdStrike-style runtime monitor for AI coding agents — network, filesystem, process visibility + security alerts

23
2
100% credibility
Found Mar 09, 2026 at 10 stars 2x -- GitGems finds repos before they trend. Get early access to the next one.
Sign Up Free
AI Analysis
Python
AI Summary

This open-source tool provides a dashboard to monitor AI coding agents' activity on your machine, including network connections, file access, resource use, token costs, and sensitive data exposure.

How It Works

1
🔍 Hear About the Watcher

You learn about a friendly tool that keeps an eye on AI helpers like Claude or ChatGPT while they work on your computer.

2
📦 Easy Setup

Install it quickly with one simple command, no tech skills needed.

3
🚀 Start Watching

Turn on the monitor and open your private dashboard in your web browser.

4
👀 See It All Live

Run your AI coding tools normally and watch in real-time what files they touch, connections they make, and costs they rack up.

5
⚠️ Spot Risks

Get instant alerts if sensitive info like passwords or keys slips into AI chats, ranked by danger level.

✅ Full Control Gained

Enjoy peace of mind with charts on usage, timelines of activity, and security checks, knowing exactly what your AIs are doing.

Sign up to see the full architecture

4 more

Sign Up Free

Star Growth

See how this repo grew from 10 to 23 stars Sign Up Free
Repurpose This Repo

Repurpose is a Pro feature

Generate ready-to-use prompts for X threads, LinkedIn posts, blog posts, YouTube scripts, and more -- with full repo context baked in.

Unlock Repurpose
AI-Generated Review

What is ai-runtime-monitor?

This Python tool delivers CrowdStrike-style runtime monitoring for AI coding agents like Claude Code, Cursor, and Copilot, giving full visibility into network calls, filesystem access, process activity, token spend, and sensitive data exposure. Install via pip, run `ai-monitor --start` for a zero-config dashboard at localhost:9081, and get real-time feeds, session replays, and security alerts ranked by severity. It solves the black-box problem of what agents are doing on your machine, from file reads/writes to API endpoints.

Why is it gaining traction?

Unlike basic process monitors, it unifies CLI agents, browser sessions (ChatGPT, Gemini), and deep API interception via an optional mitmproxy setup with `claude-watch --start`, capturing prompts, tool calls, and costs without config tweaks. Developers hook on the drill-down cockpit for turn-by-turn analysis, burn rate forecasts, and instant alerts for leaked AWS keys or GitHub tokens. The live dashboard tabs—Session Explorer, Alerts, Analytics—make auditing agent behavior addictive.

Who should use this?

Security-conscious devs wielding AI agents for coding tasks, like backend engineers debugging Claude's file edits or frontend folks tracking Copilot's network pings. Ideal for teams enforcing DLP on shared machines or solo hackers optimizing token budgets in Aider/Cursor workflows. Skip if you're not running agent-heavy setups.

Verdict

Promising early beta (10 stars, 1.0% credibility) with solid docs and PyPI packaging, but low maturity means expect rough edges—test in non-prod first. Grab it if agent observability gaps bug you; pair with `claude-watch` for full power.

(198 words)

Sign up to read the full AI review Sign Up Free

Similar repos coming soon.