Uniprof
Profile CPU hot paths with uniprof from your agent before release or when production feels slow.
Overview
io.github.indragiek/uniprof is a MCP server for the Ship phase that exposes a universal CPU profiler to coding agents over stdio npm.
What is this MCP server?
- Universal CPU profiler built for both humans and AI agents
- npm package uniprof (version 0.3.4) with stdio MCP via positional args: mcp run
- Agent-friendly profiling workflow alongside interactive use
- Fits pre-launch performance passes and post-deploy investigation
- Open-source project at github.com/indragiek/uniprof
- Published server version 0.3.4 on npm identifier uniprof
- 2 documented positional package arguments: mcp, run
- stdio transport per server manifest
Community signal: 404 GitHub stars.
What problem does it solve?
You suspect CPU bottlenecks but switching between terminal profilers and chat breaks flow, so hot paths stay hidden until users complain.
Who is it for?
Solo backend or systems builders who want agent-assisted CPU profiling before launch or during performance incidents.
Skip if: Teams that only need memory leak tooling, frontend-only Lighthouse checks, or profiling without local process access.
What do I get? / Deliverables
With uniprof registered, your agent can run CPU profiling workflows so you ship with evidence-based performance fixes instead of guesses.
- Stdio MCP server launching uniprof via mcp run
- CPU profile data your agent can help interpret for optimization
- Repeatable local profiling setup tied to your IDE agent
Recommended MCP Servers
Journey fit
CPU profiling is most critical when you are hardening performance before ship and when diagnosing regressions in operate. uniprof is a universal CPU profiler—canonical shelf is ship → perf as the primary home for latency and hot-spot work.
How it compares
CPU profiler MCP for agents, not a dependency vulnerability scanner or cloud APM dashboard.
Common Questions / FAQ
Who is io.github.indragiek/uniprof for?
Developers and agents who need to measure and explain CPU usage on local or dev workloads, especially indie builders optimizing backends and CLIs.
When should I use io.github.indragiek/uniprof?
Use it during ship perf hardening, when operate infra shows high CPU, or while building CPU-sensitive backend code you want to profile early.
How do I add io.github.indragiek/uniprof to my agent?
Install the uniprof npm package, configure your MCP client to run uniprof with stdio and positional arguments mcp and run, then reconnect your agent.