All skills
data-middleware-change-builder
ProfessionalRuntime top levelUse analysis-agent or task-agent for database, cache, queue, search, consistency, or source-of-truth work; skip unrelated work.
- Group
- Implementation owners
- Routing mode
- automatic
- Supported roles
- MainAnalysisTaskReview
- Source
- SKILL.md
Use when
Trigger signals that make this skill the right owner
- database cache queue or search change
- consistency risk
Do not use when
Anti-triggers — as binding as the triggers above
- no middleware impact
- unrelated source inspection
Required inputs
What must be supplied before this skill can decide anything
- source of truth evidence
- failure requirements
Output contract
What this skill owes the next role
- state-ownership decision
- stateful boundary changes
- recovery risk
Escalation signals
When this skill must hand the decision back
- scope, authority, or material risk exceeds the selected task contract
Related expertise
21 Layer 3 candidates may be loaded behind this skill. A single task loads at most three, and only when the current evidence justifies them.
- data-migration-design
- transaction-consistency
- concurrency-control
- idempotency-retry-design
- cache-design
- permission-boundary-modeling
- observability
- performance-budgeting
- relational-database
- indexing-query-optimization
- sql-professional-usage
- regression-testing
- release-rollback
- ai-product-extension
- bigdata-product-extension
- file-storage-processing
- nosql-database
- search-analytics-design
- distributed-workflow-consistency
- configuration-runtime-policy
- dependency-vulnerability-scanning
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 | A bounded L2 review needs checks for source of truth, consistency, access patterns, indexes, cache, queues, search, storage, migrations, lifecycle, observability, and evidence | The inline quality gate is enough or detailed evidence/recovery patterns are required | AnalysisTask |
| evidence-patterns.mdevidence-pattern | Closure depends on query-plan artifacts, cache tests, queue/replay proof, migration/backfill reports, search cutover evidence, freshness, or proof limits | No data correctness claim depends on runtime evidence or the body evidence contract is sufficient | AnalysisTask |
| index.mdindex | competing data middleware change builder references require dependency, conflict, or output-fragment selection | the data middleware change builder root or a task-named reference already resolves selection | AnalysisTask |
| recovery-patterns.mdbenchmark-pattern | Failure recovery, replay, rollback, reconciliation, release watch, or owner/trigger/stop condition is material to correctness | The change has no material recovery path beyond normal tests and the body quality gate is enough | AnalysisTask |