Loading ToolBench
An index-free web search engine: an LLM recalls where the answer lives, liberdex fetches and ranks it. Provides HTTP API, search page, and MCP endpoint.
Two tools with detailed parameter schemas and descriptions, but significant gaps in output documentation and error handling guidance. The 'search' tool has a comprehensive input schema (14 parameters with types, enums, and descriptions), exceeding the baseline of 4 params/tool. The 'extract' tool is simpler (4 params). Both descriptions are adequate (60-80 chars), meeting the 10-1024 char baseline. However, neither tool documents its output schema structure, forcing LLMs to infer response fields. Error handling is absent, no guidance on retryability, user-fixable errors, or recovery paths. Parameter naming follows verb_noun convention (search, extract), but lacks natural-identifier support (e.g., search requires a raw query string; no intent-based shortcuts). No tool annotations (readOnlyHint, destructiveHint) despite both being read-only operations.
Extract structured data from web pages using a provided JSON Schema.
Search the web using liberdex's index-free engine. Returns ranked results with optional answer synthesis.
Output schemas not documented. LLMs cannot infer response structure (fields, types, pagination). Both tools return complex objects (SearchReply, ExtractReply) but descriptions do not specify what fields to expect or how to chain results.
No error handling guidance. Tools provide no recovery hints (e.g., 'If search fails, try reducing budget or depth'). LLMs receive raw errors with no actionable next steps.
Missing tool annotations. Both tools are read-only (no side effects), but lack readOnlyHint annotation. This prevents clients from optimizing caching or parallelization.
Parameter 'budget' in search tool lacks validation constraints. Description says '0.5-60.0 seconds' but no min/max bounds in schema. LLMs may pass invalid values (negative, >60).
Inferred effective spec: 2026-07-28+.
| Scored | Grade | Overall | Spec posture | Rubric |
|---|---|---|---|---|
| 2026-09-23 | D | 56 | 2026-07-28+ | v2 |
Extract tool's 'schema' parameter accepts arbitrary JSON Schema but provides no guidance on supported features, complexity limits, or error handling for invalid schemas.