route_trace read
Trace a route to its indexed handler, matching files, and nearby evidence.
Input
{ "route": "/api/manager/onboarding", "verb": "POST" }route_context read
Route resolution plus surrounding layout, middleware, and client/server boundary checks.
trace_file read
Trace a file through symbols, imports, dependents, routes, and related evidence.
Useful as a "explain this file" entry point.
trace_edge read
Trace a handler or edge function through routes, callers, tables, RPCs, and DB triggers.
trace_error read
Trace an error term across throw sites, catch handlers, and PL/pgSQL bodies.
First call when debugging a runtime error you've seen in Sentry/logs.
trace_table read
Trace a table through schema shape, RPC-to-table edges, and app-code .from() call sites.
trace_rpc read
Trace an RPC through definition, body references, table refs, and app-code .rpc() call sites.
flow_map read
Convert a graph path into ordered steps and boundaries — useful when explaining a multi-hop request flow to the agent.
graph_path read
Find a shortest typed path between two entities in the graph (file → table, route → RPC, etc.).
graph_neighbors read
Adjacent entities for a node — imports, exports, callers, callees.
auth_path read
Trace likely auth boundaries for a route, file, or feature.
Returns the route → likely auth-guard files → role/session call sites → relevant tables. The first call when the question is "is this protected?".
table_neighborhood read
Table-centered context bundle: schema, FKs, RLS, RPCs that touch it, app-code call sites.
rpc_neighborhood read
RPC-centered context bundle: signature, body, tables touched, app-code callers.