agent-sh-perf

Plugin agent-sh-perf from marketplace cascade-code-testing-misc in joshuarweaver/cascade-code-testing-misc. Commands read from the repo's manifest — 1 month ago.

Skills (8)

• perf-analyzer

  Use when synthesizing /perf findings into evidence-backed recommendations and a continue or stop decision.

• perf-baseline-manager

  Use when storing, comparing, or consolidating /perf baselines. Keeps exactly one baseline JSON per version.

• perf-benchmarker

  Use when running performance benchmarks, establishing baselines, or validating a regression with sequential runs. Keeps runs sequential and long enough to trust.

• perf-code-paths

  Use when mapping the code paths, entry points, and likely hot files for a performance scenario before profiling.

• perf-investigation-logger

  Use when appending a structured entry to a /perf investigation log: user quotes, phase summary, evidence, decisions.

• perf-profiler

  Use when profiling CPU or memory hot paths, generating flame graphs, or capturing JFR or perf evidence for a /perf scenario.

• perf-theory-gatherer

  Use when generating performance hypotheses for a scenario, backed by git history, measurements, and code evidence.

• perf-theory-tester

  Use when running a controlled experiment to test one /perf hypothesis: single change, repeated runs, revert.

Slash commands (1)

• /agent-sh-perf:perf

  Structured performance investigation with baselines, profiling, and evidence-backed decisions

Subagents (6)

• perf-analyzer

  Synthesize a /perf investigation (baseline, breaking point, constraints, profiles, experiments) into evidence-backed recommendations and a continue or stop recommendation.

• perf-code-paths

  Map the code paths, entry points, and likely hot files for a /perf scenario before profiling.

• perf-investigation-logger

  Append a structured entry to a /perf investigation log with exact user quotes, evidence pointers, and decisions.

• perf-orchestrator

  Run a /perf investigation end to end as a subagent: phases, measurements, delegation to the perf skills, and the investigation log. Use when a caller hands off a whole performance investigation.

• perf-theory-gatherer

  Generate up to 5 evidence-backed performance hypotheses for a /perf scenario after reading git history and the measurements so far.

• perf-theory-tester

  Test one /perf hypothesis with a controlled experiment: one change, repeated sequential benchmark runs, then revert to the clean baseline.

Similar plugins