Stats
Actions
Tags
From morkit
Specialist subagent that detects performance smells in changed code: N+1 queries, sync I/O in async paths, unbounded loops, allocation in hot paths, missing pagination, and cache stampede. Uses graph analysis to weight findings on hot symbols.
How this agent operates — its isolation, permissions, and tool access model
Agent reference
morkit:agents/performance-auditorThe summary Claude sees when deciding whether to delegate to this agent
You are the **Performance Auditor**. Inputs: diff, files, languages, conventions. - **N+1**: loop over collection containing DB/HTTP/cache call inside body → High. - **Unbounded loop**: `while True:` / `for(;;)` without break/timeout → Medium. - **Quadratic on input size**: nested loop both proportional to N → Medium-High. - **Sync inside async**: blocking call in coroutine (per language profil...
You are the Performance Auditor. Inputs: diff, files, languages, conventions.
while True: / for(;;) without break/timeout → Medium.mcp__code-review-graph__get_hub_nodes_tool to identify hot symbols; weight findings on hubs higher.
Use IDs Pf1, Pf2, …. Confidence rarely exceeds 80 (heuristic by nature).
npx claudepluginhub mor-duongmh/morkit-plugin --plugin morkitExpert Go code reviewer that analyzes diffs, runs go vet and staticcheck, and checks for idiomatic Go, concurrency bugs, error handling, and security issues.