breeze

Inventory ALL entry points in a backend repo — REST routes, GraphQL operations, and queue/event/cron handlers — plus orphan handlers, then write the full inventory to entrypoints.json on disk and return a single compact summary line (counts + path + flags). Designed to be invoked ONCE by the generate-functional-from-backend skill so the parent's context stays lean. Does NOT do user-facing gates (GraphQL confirmation, exclusions) — it flags GraphQL operations as needs_confirmation and leaves those decisions to the parent. Does NOT build the functional graph or upsert — that is the per-EP backend-flow-structuring-agent's job.

Take ONE backend entry point (REST route, GraphQL operation, or queue/event/cron handler) plus its mechanical persona (System or External System), read the handler and every constructor-injected service/repository/client, produce a complete Functional Graph subtree (Persona → Outcomes → Scenarios → Steps → Actions) byte-valid against the upsert schema, self-validate it (schema / rule-a / chain / persona / citations / side-effect coverage / polymorphic split), write it to disk, and POST it to the Breeze /functional-graph/v2/upsert REST endpoint. Designed to be invoked by the generate-functional-from-backend skill (one call per entry point). Returns a single summary line with HTTP status and functionalId.

Take ONE frontend or backend entry point plus the persona that owns it, read the relevant code, produce a complete Functional Graph subtree (Persona → Outcomes → Scenarios → Steps → Actions) byte-valid against the upsert schema, self-validate it (schema / rule-a / forbidden words / citations), write it to disk, and POST it to the Breeze /functional-graph/v2/upsert REST endpoint. Designed to be invoked by the generate-functional-from-ui skill (one call per (EP, persona) pair). Returns a single summary line with HTTP status and functionalId.

Create or update nodes in the functional graph (Persona > Outcome > Scenario > Step > Action) from code, documents, or Figma designs. Use when: "update functional graph", "add to graph", "capture this in the functional graph", after analyzing a Figma frame, document, or code cluster.

Generate design graph nodes (UserJourney, Flow, Page, Component) from functional graph. Maps Scenario→UserJourney, Step→Flow/Page, Action→Component. Use when: "create design from functional", "generate UI structure", "map functional to design graph".

Search the functional, design, code, and architecture graphs to answer questions about the system. Smart-routes to one graph, a subset, or all of them based on query intent. Use for: feature discovery, "how does X work", "who handles Y", finding code or UI implementations, schema lookups, and cross-layer questions. Default entry point for any question about the project.

Analyze a requirement (from Jira, document, image, diagram, or text) against the existing Architecture Graph. Maps the requirement to the 8-layer model, runs impact analysis via the Code Graph, anchors user-facing components to Functional Scenarios, detects reuse opportunities and gaps, and writes a structured analysis report back to the Jira ticket. The Architecture Graph persists only the 8 layers — all analysis output is ephemeral and lives in the write-back.

Compare Figma design system components against the Breeze design graph component registry. First builds existingcomponents.json by iteratively fetching all Component nodes from MCP, then reads a user-provided figmaComponents.json file and performs a multi-dimensional comparison: name matching (with semantic equivalence resolution), type (ATOM/MOLECULE/ORGANISM/TEMPLATE) alignment, and supportingComponents parity. Outputs a structured comparison report. Use when: "compare components", "figma vs design graph", "component drift analysis", "compare figma components", "component comparison", "design system audit".