القائمة
Analyze a Go heap profile (pprof). Runs go tool pprof to extract top consumers, call graphs, and actionable findings. Requires a .pb.gz file — use golang-pprof-collect to obtain one.
التثبيت باستخدام Codex أو Claude انسخ هذا Prompt والصقه في Codex أو Claude أو مساعد آخر ليراجع صفحة Skill ويثبّتها لك.
استنادا إلى تصنيف SOC المهني
Collect a Go pprof profile from a running process. Handles Kubernetes port-forwarding, endpoint discovery, and file validation. Supports all pprof profile types (cpu, heap, goroutine, mutex, block, trace, etc.).
Analyze a Go CPU profile (pprof). Runs go tool pprof to extract hot functions, call graphs, and actionable findings. Requires a .pb.gz file — use golang-pprof-collect to obtain one.
Analyze a Go goroutine profile (pprof). Extracts goroutine counts, blocking stacks, and leak indicators. Requires a .pb.gz and/or debug=2 text dump — use golang-pprof-collect to obtain them.
| name | golang-profile-heap |
| description | Analyze a Go heap profile (pprof). Runs go tool pprof to extract top consumers, call graphs, and actionable findings. Requires a .pb.gz file — use golang-pprof-collect to obtain one. |
| version | 1.0.0 |
| category | performance-analysis |
| tags | ["go","pprof","heap","memory","gc-pressure"] |
| tools | ["go (go tool pprof)"] |
| inputs | ["Heap profile file (.pb.gz)"] |
| outputs | ["Top memory consumers (inuse_space, inuse_objects, alloc_space)","Hot allocation paths and GC pressure indicators","Call chain analysis","Actionable optimization recommendations"] |
| depends_on | ["golang-pprof-collect (for profile collection)"] |
Ask the user for the absolute path to a heap profile file (.pb.gz).
If they don't have one, run the golang-pprof-collect skill with profile type heap to collect it first, then return here with the file path.
Run these go tool pprof commands against the .pb.gz file. Capture all output for analysis.
Top by flat inuse_space (default):
// turbo
go tool pprof -top -inuse_space <profile_path>
Top by cumulative inuse_space:
// turbo
go tool pprof -top -cum -inuse_space <profile_path>
Top by inuse_objects (object count pressure):
// turbo
go tool pprof -top -inuse_objects <profile_path>
Top by alloc_space (total allocations over lifetime):
// turbo
go tool pprof -top -alloc_space <profile_path>
Text call graph for the top flat consumer: After identifying the top function from flat inuse_space, run:
// turbo
go tool pprof -text -focus=<top_function_package_prefix> -inuse_space <profile_path>
Produce a structured analysis with these sections:
Table: rank, function, flat MB, flat%, cum MB, cum%
Identify functions with high alloc_space but low inuse_space — these indicate high churn / GC pressure.
Flag functions holding large numbers of small objects — these stress the GC scanner.
For the top 3-5 consumers, trace the cumulative call chain and identify:
For each major consumer, suggest:
Suggest any additional pprof commands that would help (e.g., -focus, -ignore, diff between two profiles, flamegraph via go tool pprof -http=:9090).