Go pprof Web

The tool analyzes Go pprof web profiling data to expose performance characteristics of Go applications. It targets developers, performance engineers, and SREs who need repeatable, machine-friendly analysis from CPU, memory, and blocking profiles. Use cases include identifying CPU hotspots in a web service, memory churn in batch jobs, and latency-sensitive goroutine contention. It is designed to produce structured outputs suitable for automation, reporting, or feeding AI agents.
The tool implements a domain-focused parsing and aggregation pipeline for pprof data. Required features include: ingesting profile data from local files or remote URLs; parsing CPU, memory, and blocking profiles; normalizing units (ns for time; bytes for memory); computing per-function metrics (total time, samples, allocations); building a lightweight call-graph representation with parent-child contributions; producing summary metrics and structured data for downstream tooling. Optional/advanced features include: filtering by function name, time window, or sampling rate; exporting results as JSON or CSV; supporting multiple profiles in a single run; and detecting outliers.
Inputs include profile_type (CPU, memory, block, mutex), data_source (path or URL), time_window (start-end), and optional filters (focus_function, ignore_noise). Outputs include a structured report with total_samples, total_time_ns, top_functions by time, allocation bytes, and a call-graph data set; timeseries-ready data can be consumed by charts. Validation ensures profile format compatibility and unit consistency.
Parses pprof proto/binary; aggregates samples by function; sums self-time and cumulated time; computes percentages of totals; ranks top contributors; derives simple call paths; outputs JSON-friendly structures for automation. Edge cases and external data failures are handled gracefully. The tool assumes input is valid Go pprof data or a supported dump; excludes UI rendering, editing of profiling data, and non-Go profiler formats. It is not a replacement for Go toolchain installations.