Utility skills reference
Utility skills provide supporting capabilities for requirements changes, security auditing, traceability verification, and project observability.req-change
Manages requirements changes with full pipeline cascade.Invocation
Change types
New capability, feature, or constraintExample: “Necesitamos soporte para bulk export de CVs”
Change to existing requirement behavior, threshold, or scopeExample: “El timeout de extraccion debe ser 480s en vez de 360s”
Remove or sunset existing requirementExample: “Ya no necesitamos el matching por percentil”
Maintenance categories (ISO 14764)
Bug fix, incorrect behaviorUrgency levels: P0 (Critical), P1 (High), P2 (Medium), P3 (Low)
External API change, dependency update, platform migration
Feature enhancement, new capability
Tech debt reduction, code health improvement
Execution phases
Inventory & context loadingGlobs spec files, reads REQUIREMENTS.md, glossary, CLARIFICATIONS.md, AUDIT-BASELINE.md, pipeline-state.json. Builds traceability index.
Change request intake & classificationParses input, classifies by type (ADD/MODIFY/DEPRECATE), maintenance category (ISO 14764), urgency (for corrective). Applies SWEBOK 5-Whys for unclear requests.
Impact analysisIdentifies direct impact (MUST change) and indirect impact (SHOULD review). Traces full chain per affected REQ. Detects conflicts (contradicts INVs, ADRs, BDD, RNs). Rates complexity (Low/Medium/High/Very High). Analyzes commit impact via git (if available). Enriches with SDD MCP server data (if available).
Clarification & decisionEvaluates against 7 SWEBOK properties (Unambiguous, Testable, Binding, Atomic, True, Stakeholder vocabulary, Acceptable). Presents questions with options and recommendations. Collects answers. Generates clarification log.
Change plan generationFor ADD: generates new REQ with full template, drafts new UC/API/BDD/INV/ADR/RN. For MODIFY: shows before/after of every affected section. For DEPRECATE: marks REQ as Deprecated, generates migration/sunset plan. Applies Atomic Cross-Check Rule. Groups changes by document.
User review & approvalBatch mode (default with
--batch): apply all. Interactive mode: approve/modify/reject per CR.Change executionApplies changes in order: REQUIREMENTS.md first (source of change), then domain/, invariants, use-cases, workflows, contracts, tests, nfr, adr, runbooks, CLARIFICATIONS.md, CHANGELOG.md. Creates git commits.
Alignment auditFocused mini-audit on affected documents: REQ → Spec Forward Coverage, Spec → REQ Backward Coverage, Traceability Complete, Cross-Refs Valid, Glossary Compliance, Contradiction Check, EARS Compliance, Testability. Auto-fixes if possible.
Change report generationGenerates comprehensive report with: executive summary, changes applied (new/modified/deprecated), new artifacts created, impact analysis, alignment audit results, clarification decisions, context for planning/implementation, recommended next steps.
Pipeline cascade executionComputes invalidation scope (which stages need re-run). Updates
pipeline-state.json (marks stale). Executes based on cascade mode (manual/dry-run/plan-only/auto). Generates cascade report (if auto/plan-only).Cascade modes
Update pipeline-state.json, print recommended commands
Invoke downstream skills in sequence: spec-auditor (focused), test-planner, plan-architect (regenerate FASEs), task-generator (incremental), task-implementer (new tasks only)
Compute scope, no changes to pipeline-state.json
Cascade through planning but not implementation
Output artifacts
Detailed change plan with before/after for every affected document
Comprehensive change report with full context for downstream skills
(if cascade=auto/plan-only) Execution log of cascaded skill invocations
Updated with new/modified/deprecated requirements
Updated specifications reflecting requirement changes
security-auditor
Audits specifications for security defects.Invocation
Security defect categories
Detects security-specific defects beyond general spec-auditor categories.Output artifacts
Security-specific audit report
traceability-check
Verifies traceability chain integrity.Invocation
Checks
Every REQ has at least one spec implementing it
Every spec traces back to at least one REQ
REQ → UC → WF → API → BDD → INV → ADR links all valid
No artifacts without traceability
Output artifacts
Traceability verification report with gaps and broken links
pipeline-status
Displays current pipeline state and stage status.Invocation
Output
Readspipeline-state.json and displays:
- Stage statuses (done/pending/stale)
- Last run timestamps
- Stale reasons
- Next recommended action
session-summary
Generates session activity report.Invocation
Output artifacts
Session activity reportSections:
- Skills invoked
- Artifacts created/modified
- Decisions made
- Pipeline state changes
- Time spent per phase
setup
Initializes SDD directory structure.Invocation
Created directories
Requirements documents
Technical specifications with subdirectories: domain/, use-cases/, workflows/, contracts/, tests/, nfr/, adr/, runbooks/
Implementation plans and FASE files
Task documents
Audit reports and baselines
Change requests and reports
Output artifacts
Initialized pipeline state with all stages pending
(if not exists) Project context template
sync-notion
Bidirectional synchronization with Notion databases.Invocation
Configuration
Requiressync-notion/config.json:
Sync protocol
Fetches from Notion, updates local SDD artifacts, marks conflicts
Sends local changes to Notion, creates/updates pages and database rows
Bidirectional: pulls first, resolves conflicts, then pushes
Output artifacts
Sync activity reportSections:
- Items pulled from Notion
- Items pushed to Notion
- Conflicts detected and resolution
- Sync statistics
code-index
Builds symbol-level code intelligence graph.Invocation
Indexed data
Functions, classes, interfaces, types with signatures and locations
Caller-callee relationships
Module dependencies
Code → spec traceability via
Refs: commentsOutput artifacts
Symbol table and call graph in MCP-compatible format
MCP integration
Enablessdd_context, sdd_impact, and sdd_search MCP tools for:
reconcile— enriched impact analysisreq-change— commit impact analysisreverse-engineer— code analysis
dashboard
Generates HTML visualization of project state.Invocation
Visualizations
Stage status, progress bar, next action
Interactive graph: REQ → UC → WF → API → BDD → INV
Coverage percentages, defect counts, test coverage
Dead code, tech debt, audit findings
Timeline of skill invocations and artifact changes
Output artifacts
Interactive HTML dashboard with embedded data
CSS, JavaScript, and visualization libraries