CtrlK
BlogDocsLog inGet started
Tessl Logo
PluginPublic

testland/flame-graph-analyzer

Use for any question or task about reading a flame graph or CPU profile, including finding the hot path after a slow release or a regression traced to one commit. Reads flame graphs and profiles from py-spy, async-profiler, Go pprof or clinic.js, finds the hottest frames, tells CPU-bound from lock contention or allocation pressure and suggests the next step.

Quality

91%

Does it follow best practices?

Run evals on this skill

Adds up to 20 points to the overall score

View guide
SecuritybySnyk

Passed

No findings from the security scan

Overview
Quality
Evals
Security
Files

Quality

Content

86%Weight 40%Scale 1-5

Reviews the quality of instructions and guidance provided to agents. Good implementation is clear, handles edge cases, and produces reliable results.

The body is an actionable, well-structured overview: executable commands, a worked example with real output, a signature-classification table, anti-patterns, and honest limitations, with details correctly deferred to a verified one-level-deep reference. The residual gaps are minor: the sort command and reference link are repeated, and the workflow lacks an explicit what-to-do-on-failure loop.

Suggestions

State the sort command once (in "Reading the hot path") and reference it from the worked example instead of repeating it three times; similarly, link the reference file once per section purpose rather than four times.

Add an explicit failure path to the workflow, e.g. "if the re-profiled delta doesn't close, re-classify the new top leaf and iterate" — currently the loop is only implicit.

Trim framing sentences like "This skill is language-agnostic... surfaces a hypothesis the engineer can act on"; the same information is carried by the tool list and workflow steps.

DimensionReasoningScore

Conciseness

Mostly dense, table-driven content with no explanations of concepts Claude already knows, but "sort -k2 -n -r folded.txt | head -5" appears three times, the same reference link is repeated four times, and the language-agnostic sentence is mild padding — minor trimmable instances, so anchor 4 rather than 5 or 3.

4 / 5

Actionability

Fully executable guidance: copy-paste capture and sort commands ("npx clinic flame -- node app.js", "sort -k2 -n -r folded.txt | head -5"), a realistic folded-stacks sample output, concrete frame-name signatures ("gc::scavenge", "pthread_mutex_lock"), and named per-runtime fixes — anchor 5.

5 / 5

Workflow Clarity

"How to use" gives a clean 5-step sequence with an explicit validation step ("Re-profile after the change to confirm the delta") and a quantitative expectation ("expect rank 1 to drop below 10%"), but no explicit failure-path recovery if the delta doesn't close — anchor 4 rather than 5.

4 / 5

Progressive Disclosure

SKILL.md is a true overview and the single bundle reference (references/capturing-and-interpreting-flame-graphs.md, verified to exist) is well-signaled, one level deep, and holds exactly the deferred content its section headers claim (per-runtime capture wiring, remediation catalog, worked cases) — anchor 5.

5 / 5

Total

18

/

20

Passed

Description

92%Weight 40%Scale 1-5

Based on the skill's description, can an agent find and select it at the right time? Clear, specific descriptions lead to better discovery.

A strong description: it names the domain, tools, and a complete chain of concrete actions, and opens with an explicit "Use for..." trigger clause covering realistic scenarios (slow release, bisected regression). The only weakness is slightly thin coverage of natural synonyms a user might say (e.g. "profiling", "CPU spike").

DimensionReasoningScore

Specificity

Lists multiple concrete actions covering the full workflow — "Reads flame graphs and profiles from py-spy, async-profiler, Go pprof or clinic.js", "finds the hottest frames", "tells CPU-bound from lock contention or allocation pressure and suggests the next step" — with no coverage gap within its niche, matching the comprehensive anchor rather than anchor 4.

5 / 5

Completeness

Explicitly answers both: "Use for any question or task about reading a flame graph or CPU profile, including finding the hot path after a slow release or a regression traced to one commit" gives concrete when-triggers, and the action list gives a clear what — anchor 5, not 4, since the when-clause is fully explicit.

5 / 5

Trigger Term Quality

Good natural keywords ("flame graph", "CPU profile", "hot path", "slow release", "a regression traced to one commit") plus tool names, but common variations like "profiling", "CPU usage", or "perf" are missing, so it fits anchor 4 rather than the comprehensive-synonym coverage of anchor 5.

4 / 5

Distinctiveness Conflict Risk

Named profilers (py-spy, async-profiler, Go pprof, clinic.js) and flame-graph-specific vocabulary establish a clear niche with distinct triggers and minimal overlap risk with related performance skills; nothing bumps it down to anchor 4's "minor overlap risk".

5 / 5

Total

19

/

20

Passed

Validation

100%

Checks the skill against the spec for correct structure and formatting. All validation checks must pass before discovery and implementation can be scored.

Validation — 16 / 16 Passed

Validation for skill structure

No warnings or errors.

Reviewed

On this page