sa-skills

Backfill EXPLORER_HEADER blocks into existing docs/NN-*.md and docs/components/**/*.md files (sa-skills:architecture-explorer-headers — surfaces key_concepts, technologies, components, scope, related_adrs in the first 60 lines so the architecture-explorer agent classifies legacy docs accurately).

Release architecture version (architecture-docs Workflow 10 — Draft→Released, CHANGELOG, architecture-v{version} git tag; PR by default, --direct opts out)

Set up the sa-skills plugin in the current project — merges required permissions, enabledPlugins, and extraKnownMarketplaces from the plugin's settings.json.example into your project's .claude/settings.json (non-destructive).

Front door for architecture documentation. Two modes — (1) MANIFEST mode (no query): walks the canonical layout (ARCHITECTURE.md, docs/, adr/) and emits an EXPLORE_MANIFEST listing every doc + ADR with their <!-- EXPLORER_HEADER --> metadata; (2) FINDINGS mode (with query): runs Explore-agent-style content discovery (parallel Grep + targeted Reads with heading context) over the same surface and emits an EXPLORE_FINDINGS block. The two modes are mutually exclusive — query mode focuses on finding what was asked, no manifest noise. Tools: Read, Glob, Grep. MUST ONLY be invoked by skill orchestrators or by Claude itself for free-form architecture Q&A.

Builds per-component handoff payloads from shared architecture documentation. Reads docs/01, docs/04–09, and adr/ once on Sonnet (saving Opus tokens in the orchestrator's main context), slices each shared doc per component, indexes ADRs by term and intersects against component metadata, parses each component file into a structured YAML block, deduplicates excerpts that appear in 3+ payloads to .cache/sa-skills/handoff-payloads/_shared.md, computes SHA-256 fingerprints, and consults handoffs/.manifest.json to decide SKIP vs REGEN per component. Returns a CONTEXT_RESULT block listing each component's payload path, hash, decision, and reason. MUST ONLY be invoked by the `architecture-dev-handoff` skill orchestrator — never call directly.

Universal Compliance Contract Generator — generates any of the 10 compliance contracts from ARCHITECTURE.md. Receives contract_type in prompt to determine which domain config and template to use. MUST ONLY be invoked by the `architecture-compliance` skill orchestrator — never call directly.

Universal architecture document export generator — produces Word (.docx) deliverables for Solution Architecture (executive summary + ADRs), Component Handoff, Compliance Contract, or Security Posture (consolidated security validation checklist) workflows. Pinned to Sonnet 4.6 to prevent cost-per-run drift when teams run frequent exports. MUST ONLY be invoked by the `architecture-docs-export` skill orchestrator — never call directly.

Universal single-asset generator for Component Development Handoffs — produces ONE deliverable asset file (OpenAPI, DDL, Kubernetes manifest, AsyncAPI, CronJob, Avro/Protobuf schema, Redis key schema, or C4 descriptor) for one C4 L2 component, using a pre-sliced architecture payload. The orchestrator pins the model per Task() call (sonnet for code-style assets, haiku for descriptor-style assets). MUST ONLY be invoked by the `architecture-dev-handoff` skill orchestrator — never call directly.

Run risk and design-characteristics analyses over ARCHITECTURE.md documentation. Produces date-stamped reports in analysis/ covering ten lenses across two groups: HIGH-priority runtime/security — SPOF (single points of failure), Blast Radius (downstream cascade impact), Bottleneck (throughput chokepoints), Cost Hotspots (Pareto cost concentration), STRIDE (per-trust-boundary security threats); Strategic/sustainability — Vendor Lock-in (portability risk and exit cost), Latency Budget (per-hop SLO decomposition), Tech Debt/EOL (currency and deprecated tech), Coupling (fan-in/fan-out and cycles), Data Sensitivity (PII flow and encryption gaps). Each analysis can be requested individually, as a group, or all ten run in parallel. A consolidated Security Posture option (analysis 12) merges the STRIDE and Data Sensitivity reports into a single reviewer-fillable validation checklist of every security control to validate (markdown-only; exportable to a Word worksheet via architecture-docs-export). Output: analysis/<TYPE>-<YYYY-MM-DD>.md (default) OR analysis/<TYPE>-<YYYY-MM-DD>.html (interactive d3.js report; format is selected at runtime — Step 2.4). Requires ARCHITECTURE.md to exist (created by architecture-docs skill). Do NOT invoke for compliance contracts (use architecture-compliance), peer quality review (use architecture-peer-review), or ADR management (use architecture-definition-record).

Generate Business and Application blueprint markdown files from ARCHITECTURE.md. Extracts architecture data to fill standardized organizational templates (datos de iniciativa / initiative data). Detects architecture document language and selects the matching template automatically. Invoke when the user asks to generate blueprints, initiative data sheets, datos de iniciativa, or organizational architecture forms.

Use this skill to review compliance contract portfolio health. Validates all 10 contracts exist and are ≤6 months old, triggers regeneration for missing/expired contracts, and generates an interactive playground for exploring concept gaps across all contracts to understand what needs to be addressed to reach the auto-approve threshold (≥8.0/10).

Generate Compliance Contracts (Contratos de Adherencia) from ARCHITECTURE.md files

Use this skill to create or update docs/components/README.md — the only sanctioned way to modify the component index table. Invoke when adding, removing, or updating components, syncing the index, or migrating flat components to C4 multi-system structure. This is the ONLY skill for C4 component migration — do not use architecture-docs for component migration.