List agent tools

Returns the catalogue of tools available to the embedded agent, read live from the loop's registry. Each tool entry carries a stable name, a short description, an impact level (read, draft_mutation, live_mutation, irreversible), the capability the tool requires, its JSON Schema argument shape (input_schema), and can_invoke: whether THIS key can run it via POST /v1/agent/tools/{name}/invoke. An API key runs as the least-privilege viewer role, so the invocable set is the read subset; the rest are listed so a caller can see the full surface the in-product agent operates.