--- description: >- TRIGGER when: user asks about relationships, dependency paths, import hotspots or cycles, symbol impact, or blast radius of a proposed change. Covers `graph_neighbors`, `graph_path`, `flow_map`, `change_plan`, `imports_deps`, `imports_impact`, `imports_hotspots`, `imports_cycles`, `symbols_of`, `exports_of`. when_to_use: >- Use for relationship questions, dependency traversal, import/symbol impact, end-to-end flows, and planning multi-file changes. allowed-tools: "mcp__mako-ai__*" --- # Mako Graph Use this skill when the user's question is about relationships rather than a single entity. Graph, import, and symbol tools help estimate impact, find paths, and plan changes from indexed project facts. ## Graph Tools ### `graph_neighbors` Use to inspect immediate graph neighbors for a file, symbol, route, table, RPC, or other known node. ### `graph_path` Use to find evidence-backed paths between two known nodes. - Best when the user asks whether A reaches B or how a route connects to a table/RPC/file. - If the endpoints are ambiguous, use discovery or trace tools first. ### `flow_map` Use to map a higher-level flow across routes, components, services, tables, and RPCs. - Best for end-to-end user stories or system behavior. - Pair with `review_bundle_artifact` when a flow is being reviewed pre-ship. ### `change_plan` Use to produce an evidence-backed edit plan and expected blast radius before implementation. - Best when the user asks "what do we need to change?" or "what might break?" - Pair with artifact tools when the plan needs a durable handoff. ## Imports ### `imports_deps` Use to inspect direct imports and unresolved internal edges for one or more files. ### `imports_impact` Use to estimate downstream dependents and likely blast radius of changing a file/module. ### `imports_hotspots` Use to identify highly connected import hotspots. ### `imports_cycles` Use to detect or inspect circular dependencies. ## Symbols ### `symbols_of` Use to list important symbols declared in a file. ### `exports_of` Use to list exported symbols from a file or module. ## Feedback Logging Log `agent_feedback` when a graph, import, or symbol result here was notably useful, partial, noisy, stale, wrong, or wasted the turn. Skip routine calls. Required procedure (see `/mako-ai:mako-guide` for full rules and reason-code vocabulary): 1. Call `recall_tool_runs` to get the prior run's `requestId`. Do not fabricate one — if no run is recalled, skip feedback. 2. Call `agent_feedback` with `referencedToolName`, `referencedRequestId`, `grade: "full" | "partial" | "no"`, `reasonCodes` from the starter vocabulary in `/mako-ai:mako-guide`, and a short `reason`. ## See Also - Use `/mako-ai:mako-trace` when the target is a specific route, table, RPC, file, or error. - Use `/mako-ai:mako-neighborhoods` when one table, route, or RPC should be the center of the context bundle. - Use `/mako-ai:mako-workflow` for artifacts after a plan or flow is clear.