Loading ToolBench
Access shadcn-svelte component documentation, Bits UI API docs, and guides. IMPORTANT: This is for SVELTE components only, NOT React. Do not confuse with shadcn/ui (React version). Bits UI provides the underlying components for shadcn-svelte.
This server demonstrates solid quality across most dimensions. All 5 tools have clear action-verb names (shadcn-svelte-list, shadcn-svelte-get, shadcn-svelte-icons, shadcn-svelte-search, bits-ui-get), good descriptions (100-300+ chars each), and complete JSON Schema definitions with typed parameters and descriptions. The tool-usage prompt provides excellent meta-guidance on routing and chaining patterns. However, output schemas are not explicitly documented in the visible source code (only input schemas are shown), error handling specifics are not detailed in the tool definitions themselves, and some parameter descriptions could be more granular about constraints and expected formats. The server targets a specific niche (shadcn-svelte docs) and executes that focus well, avoiding scope creep and keeping tool semantics clean.
Bits UI primitive internals. Use only after shadcn-svelte-get returns docs.bitsuiName or tooling.bitsUi.exactName, and only when you need the underlying primitive API. Accepts canonical Bits UI names, PascalCase names, component URLs, or some shadcn component names.
PRIMARY tool for shadcn-svelte. Get docs for any component, block, chart, or doc section. Use FIRST for shadcn-svelte questions. Svelte only, not React. If the response includes tooling.bitsUi.exactName, that is the only value to pass to bits-ui-get.
Search real Lucide icons for lucide-svelte by name or tag (1600+ icons, no hallucination). Icons only — for components use shadcn-svelte-get, shadcn-svelte-search, or shadcn-svelte-list. Accepts comma- or space-separated names (e.g. 'truck, package').
Enumerate available resources: components, blocks, charts, docs, or Bits UI primitives discovered from the live websites.
Fuzzy discovery by keyword or phrase. Use when you are not sure of the exact name, or to find components to install. Typo-tolerant, returns links, install commands, and similarity scores, plus suggestions when nothing matches.
Output schemas not documented in visible source code. While input schemas are complete and well-typed, return types and response structures are not explicitly defined in the tool registration code. LLMs cannot plan downstream tool calls without knowing the response shape.
Parameter constraint documentation sparse. 'limit' and 'importLimit' in shadcn-svelte-icons have default/max hints in descriptions but no explicit min/max schema constraints. Descriptions should explicitly state ranges (e.g. 'limit 1-100, default 20') to prevent LLMs from passing invalid values.
Error handling guidance absent from tool descriptions. None of the tool definitions explain what happens on failure (e.g., 'resource not found', 'invalid package manager'). Tool descriptions should include recovery hints like 'If not found, call shadcn-svelte-list to see available resources.'
| Scored | Grade | Overall | Spec posture | Rubric |
|---|---|---|---|---|
| 2026-09-22 | B | 76 | 2026-07-28+ | v2 |
| 2026-03-09 | B | 70 | - | v1 |
Tool composition clarity. The bits-ui-get tool depends on the output of shadcn-svelte-get (via tooling.bitsUi.exactName), but this dependency is documented only in the tool-usage prompt, not in the bits-ui-get description itself. The description mentions this chaining pattern but could be more directive: 'Call shadcn-svelte-get first; if it returns tooling.bitsUi.exactName, pass that exact value to this tool.'