All skills
implementation-structure-design
FoundationJIT onlyanalysis-agent/task-agent/review-agent: use when reuse or method/class/file placement inside one owner is disputed; skip fixed placement or cross-module ownership.
- Group
- Engineering workflow
- Supported roles
- MainAnalysisTaskReview
- Source
- SKILL.md
Use when
Trigger signals that make this skill the right owner
- semantic-compatible reuse inside an established owner is disputed
- method class file or test placement inside an established owner is disputed
- co-location or private extraction inside an established owner is disputed
- owner-private new structure inside an established owner is disputed
- deliberate separate implementation inside an established owner is disputed
- generated-source separation or placement inside an established owner is disputed
Do not use when
Anti-triggers — as binding as the triggers above
- cross-module ownership or dependency direction changes
- public shared exported or cross-owner surface changes
- dependency cycle or shared package changes
- distributable SDK or library contract changes
- architecture style layer responsibilities or extension compatibility is primary
- model translation is primary
- side-effect ordering is primary
- persistence semantics is primary
- endpoint API or transport contract is primary
- behavior-preserving movement or refactoring is primary
- local flow or readability is primary
- current owner and repository convention already fix placement
Required inputs
What must be supplied before this skill can decide anything
- accepted existence behavior and owning module
- candidate owners callers tests exports generated authority and sibling patterns
- semantics authority failure lifecycle evolution and change-reason comparison
Output contract
What this skill owes the next role
- Implementation placement decision with accepted existence input, inspected reuse and deliberate-separation candidates, selected owner-private location, visibility, and co-location/extraction rationale. It records semantic and generated-source handling, tests/fixtures, evidence and validation plan, proof limits, delete/drift conditions, specialist handoffs, and residual risks.
Escalation signals
When this skill must hand the decision back
- placement would change cross-module ownership public surface dependency direction or cycles
- model effect resource persistence or endpoint semantics lack an authoritative owner
- behavior-preserving movement lacks characterization or compatibility proof
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 |
|---|---|---|---|
| object-module-decomposition.mdtargeted | co-location extraction visibility split merge or test placement remains disputed inside one owning module | the current owner and change lifecycle already fix the object or file boundary | TaskReviewAnalysis |
| reuse-and-placement.mdtargeted | repository-local reuse leaves more than one semantic-compatible owner-private placement or deliberate-separation decision | one semantic-compatible owner-private placement is already proved | TaskReviewAnalysis |
| evidence-patterns.mdevidence-pattern | placement generated-source or deliberate-separation claims need current owner consumer and freshness evidence | no placement or source-authority claim awaits closure | TaskReviewAnalysis |