Infer the type and effect row of the expression at a source location (1-based line, 1-based character column) in a .hird file.

Look up a top-level definition by name: source location, type, doc comment, and kind.

Explain a function's effect row: the canonical row plus a human-readable explanation of each effect.

Render the typed IR of one top-level definition as JSON.

Describe an actor's protocol: message constructors, state type, init, handler signatures, and the declared effect summary.

Emit the actor/effect graph rooted at an actor: reachable actors (via Send/Await/Spawn message types), supervisor relationships, and transitive tool effects.

A token-budget-aware summary of a symbol: kind, signature, effect row, doc, callers, and callees, fitted to the budget.

Approximate token costs of including the file's types, effects, actors, supervisors, tools, and function signatures in an LLM context window.