All skills
language-performance-safety
FoundationJIT onlyanalysis-agent/task-agent/review-agent: use when allocation, GC, concurrency, async/blocking, FFI, or unsafe/native behavior changes; skip when runtime safety is unaffected.
- Group
- Technology selection
- Supported roles
- MainAnalysisTaskReview
- Source
- SKILL.md
Use when
Trigger signals that make this skill the right owner
- language performance safety allocation GC async blocking FFI unsafe native hot path memory concurrency event loop runtime constraints
- CPU-bound, storage IO, network IO, file IO, blocking IO, non-blocking IO, coroutine, goroutine, thread pool, worker pool, lock held across IO, event-loop blocking, unbounded fan-out, pool sizing, per-operation client construction, response body leak, object allocation regression, design pattern performance risk
Do not use when
Anti-triggers — as binding as the triggers above
- no task-local language performance safety decision is required
Required inputs
What must be supplied before this skill can decide anything
- current task contract
- selected primary Professional Skill
- task-local trigger evidence
Output contract
What this skill owes the next role
- runtime performance and safety decision with workload evidence, allocation and scheduler boundaries, growth limits, lifecycle ownership, concurrency proof, unsafe contract, and proof limits
Escalation signals
When this skill must hand the decision back
- Runtime specific performance and safety issues often appear only under representative load or native boundary stress
Loaded by
This skill never owns a task. These professionals may pull it in as Layer 3.
Targeted references
Each reference carries its own load condition. None of them enter context by default.
| Reference | Load when | Do not load when | Required by |
|---|---|---|---|
| benchmarks-and-patterns.mdbenchmark-pattern | profiling exposes competing allocation concurrency pooling or cleanup remedies | measurement and runtime constraints select one safe remedy | TaskReviewAnalysis |
| checklist.mddecision-checklist | hot paths alter allocation bounds async blocking cleanup or unsafe code | change touches no measured performance or runtime-safety boundary | TaskReviewAnalysis |
| evidence-patterns.mdevidence-pattern | performance race cleanup or FFI claims need fresh runtime proof | current profiles stress checks and source evidence prove each claim | TaskReviewAnalysis |