All skills
backend-change-builder
ProfessionalRuntime top levelUse task-agent for bounded backend service, API, worker, or repair changes, loading authorization, consistency, retry, contract, and rollout guidance only when triggered. Skip frontend-only and read-only work.
- Group
- Implementation owners
- Routing mode
- automatic
- Supported roles
- MainAnalysisTaskReview
- Source
- SKILL.md
Use when
Trigger signals that make this skill the right owner
- backend behavior change
- service or worker change
Do not use when
Anti-triggers — as binding as the triggers above
- frontend only
- source inspection only
Required inputs
What must be supplied before this skill can decide anything
- domain invariants
- authorization and failure-contract evidence
- transaction delivery and recovery constraints
Output contract
What this skill owes the next role
- changed backend invariants and boundaries
- authorization consistency delivery and recovery decisions
- post-edit validation result and proof limits
- residual operational risk
Escalation signals
When this skill must hand the decision back
- scope, authority, or material risk exceeds the selected task contract
Related expertise
37 Layer 3 candidates may be loaded behind this skill. A single task loads at most three, and only when the current evidence justifies them.
- service-business-logic
- domain-object-identification
- transaction-consistency
- concurrency-control
- idempotency-retry-design
- permission-boundary-modeling
- authentication-authorization
- regression-testing
- logging-error-handling
- message-queue-design
- observability
- failure-diagnosis
- failure-contract-design
- implementation-structure-design
- design-pattern-selection
- minimal-correct-implementation
- go-professional-usage
- java-jvm-professional-usage
- kotlin-professional-usage
- csharp-dotnet-professional-usage
- python-professional-usage
- language-idiom-enforcement
- language-performance-safety
- ai-product-extension
- payment-trading-extension
- low-level-systems-extension
- cloud-platform-extension
- windows-platform-extension
- async-job-design
- controller-api-implementation
- dependency-wiring-lifecycle
- domain-logic-implementation
- repository-persistence
- nodejs-runtime-professional-usage
- filesystem-process-safety
- 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 |
|---|---|---|---|
| backend-output-and-gates.mdtargeted | extended fields apply when implementation closure depends on edit-to-validation ordering or repair proof | The root contract is sufficient for bounded implementation | Task |
| checklist.mddecision-checklist | A bounded implementation needs quick checks for its triggered backend risks | The root contract is enough or extended proof fields are needed | Task |
| index.mdindex | competing backend change builder references require dependency, conflict, or output-fragment selection | the backend change builder root or a task-named reference already resolves selection | Task |
| proactive-triggers.mdtargeted | Authorization, tenancy, retries, async, public contracts, transactions, irreversible mutation, or AI-generated backend code are material | No hidden backend escalator is present | Task |
| professional-modes.mdmode-contract | L3+ implementation or accepted-finding repair needs mode-specific proof and ownership limits | The root contract determines implementation evidence or the task is still diagnosis/independent review | Task |
| solution-optimality.mdtargeted | An accepted implementation chooses a material algorithm, query pattern, cache, batch/stream boundary, concurrency control, pool, queue, or other resource-sensitive design | No runtime/resource tradeoff is material or current system evidence already fixes the mechanism | Task |