Browser DevTools Network MCP
An MCP server that captures a tab's full network log (requests, responses, timing, performance traces) and replays it for an agent to debug 'why is this page slow' without needing screenshots or human narration.
- Target market
- Frontend devs and full-stack indie hackers debugging performance issues in production web apps with Claude Code or Cursor as their reasoning layer.
Problem snapshot
What this solves
The existing Chrome DevTools MCP from the CDP repo handles basic page interaction but does not expose performance traces or full network logs in a replayable form. So when a user complains 'the dashboard takes 8 seconds to load', the developer has to either screen-share with Claude or paste truncated HAR snippets, neither of which lets the agent reason about…
Full ProvenTools analysis
Unlock the full analysis and build prompt
Unlock the solution, revenue model, feature scope, technical approach, user flow, and 13-section AI build prompt.
Already have access? Sign in
Related ideas
Explore similar problems
Agent Canary Task Monitor
Runs a handful of known-good canary tasks against your live agent on a schedule and alerts the moment its output, tool path, latency, or cost drifts from baseline.
Agent Context Drift Alarm
Scans the assembled context (retrieved chunks, memory, task) for contradictions and topic mismatch before the agent answers, so it clarifies or prefers the authoritative source instead of confidently resolving a confl...
Agent Evaluation and Benchmarking Platform
A platform for systematically evaluating AI agent performance against custom test suites.