perfetto-mcp
perfetto-mcp is a local Model Context Protocol (MCP) server that turns natural-language prompts into Perfetto trace analyses (including generating PerfettoSQL) for tasks like jank/ANR detection, CPU hotspot analysis, lock contention, Binder transaction profiling, and memory leak signals.
Score Breakdown
⚙ Agent Friendliness
🔒 Security
No network interface or TLS/auth is described; the server runs locally via stdio. This reduces remote attack surface, but increases risk from any tools that may read local trace files (sensitive data). No explicit guidance on sandboxing, trace file validation, logging of sensitive data, or handling of malicious trace contents is provided. Dependencies include mcp[cli], perfetto, protobuf<5, which may warrant dependency/CVE review before use in sensitive environments.
⚡ Reliability
Best When
You have one or more Perfetto trace files locally and want an agent to orchestrate analysis by calling MCP tools that return structured JSON results.
Avoid When
You need authentication/authorization controls, network-access APIs, or strict compliance claims without additional vetting; also avoid untrusted trace content ingestion in sensitive environments without sandboxing.
Use Cases
- • Analyze Android Perfetto traces via natural-language requests
- • Diagnose jank (missing frame deadlines) and summarize frame performance
- • Detect and investigate ANR events and likely root causes
- • Identify CPU hot threads/scheduling hotspots
- • Find lock contention / thread synchronization bottlenecks
- • Profile Binder IPC latency and group slow transactions
- • Detect suspected memory leak patterns and heap dominators
Not For
- • Using as a hosted SaaS/API service (it appears intended to run locally as an MCP stdio server)
- • Production workflows requiring guaranteed, reproducible analysis without validation of generated SQL
- • Scenarios where you cannot provide local trace files (requires trace_path as input)
Interface
Authentication
README describes running the MCP server locally (e.g., uvx/uv) and provides required tool parameters (trace_path, process_name). No user authentication mechanism is documented, suggesting it runs without auth in a local developer environment.
Pricing
No pricing information; appears to be an open-source package executed locally.
Agent Metadata
Known Gotchas
- ⚠ Tool inputs require absolute trace_path and a process_name; incorrect paths/process names may yield empty or failed analyses.
- ⚠ Generated SQL/tool behavior depends on the trace containing the relevant events (e.g., ANRs); prompts should constrain time ranges and expectations.
- ⚠ Because it is a local MCP stdio server, agents need to launch it correctly (command/args) per client (Cursor/Claude/Copilot/etc.).
Alternatives
Full Evaluation Report
Comprehensive deep-dive: security analysis, reliability audit, agent experience review, cost modeling, competitive positioning, and improvement roadmap for perfetto-mcp.
AI-powered analysis · PDF + markdown · Delivered within 30 minutes
Package Brief
Quick verdict, integration guide, cost projections, gotchas with workarounds, and alternatives comparison.
Delivered within 10 minutes
Score Monitoring
Get alerted when this package's AF, security, or reliability scores change significantly. Stay ahead of regressions.
Continuous monitoring
Scores are editorial opinions as of 2026-03-30.