All skills
ai-product-extension
DomainJIT onlyFor analysis/task/review agents using a Professional Skill on models, RAG, agents, evaluation, or safety; not for work without AI decision impact.
- Routing mode
- modifier-only
- Supported roles
- MainAnalysisTaskReview
- Source
- SKILL.md
Use when
Trigger signals that make this skill the right owner
- model
- prompt
- retrieval
- embedding
- evaluation
- AI data
Do not use when
Anti-triggers — as binding as the triggers above
- deterministic product behavior with no AI surface
- AI terminology, static algorithms, or ordinary search without a model decision
Required inputs
What must be supplied before this skill can decide anything
- model and data contract, evaluation target, and safety boundary
Output contract
What this skill owes the next role
- AI-specific risks, evaluation, fallback, and observability obligations
Boundary signals
Signals that mark the edge of this skill's authority
- permission
- model context
- delegated authority
- consequential
Escalation signals
When this skill must hand the decision back
- domain risk changes public behavior, irreversible state, safety, security, or financial correctness
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 |
|---|---|---|---|
| checklist.mddecision-checklist | model prompt retrieval embedding tool or evaluation behavior crosses product surfaces | AI terminology describes a static algorithm or ordinary search without a model decision | AnalysisTaskReview |