get_type_info
Return types, parameter types, documentation via LSP hover. 50-200 tokens instead of reading entire files.
What it does
Agents often answer simple type questions in a very expensive way: search for the symbol, open the file, read surrounding lines, infer the type manually.
get_type_info short-circuits that pattern by retrieving resolved type information directly when the language server can provide it. The equivalent of hovering your mouse over a symbol in VS Code, but available to AI agents.
Why this matters
This is one of the clearest examples of CodeSift changing what an agent has to do, not just making the same thing smaller.
The native fallback for type discovery is usually “read more code.” The LSP-backed path is “ask for the resolved type.”
The benchmark framing here should emphasize fewer follow-up reads and lower cognitive noise for the model — not a rigid 1:1 claim against a single shell command.
When to use
- “What does this function return?”
- “What type does this parameter expect?”
- “What is the resolved type of this symbol?”
- Any time you’d read a file just to find a type annotation
Related tools
Benchmark note
This benchmark compares CodeSift against the closest practical native workflow an agent would use for the same task.
For some tools, that baseline is a direct shell equivalent such as rg or find.
For AST-aware, graph-aware, and LSP-backed tools, the baseline is a multi-step workflow rather than a strictly identical command.
Results should be read as agent-workflow comparisons: token cost, call count, and practical context efficiency.