Loading ToolBench

Reading the current corpus…

microfish MCP server · C · ToolBench

← Browse

microfish

github.com/vvtommy/microfish

A restricted TinyFish MCP gateway for free search and fetch capabilities.

2026-07-28+STREAMABLE_HTTP4 tools

63/100

Definition Quality77

Protocol Readiness78

Supportability31

Spec posture

Inferred effective spec 2026-07-28+·current baseline 2026-07-28

Static source inference · medium confidence · evidence: stateless requests, Streamable HTTP

Current-spec patterns detected

Summary

Four tools with complete input schemas and descriptions. All tools are read-only with clear risk classification. Naming follows verb_noun pattern (search, fetch_content, get_*, list_*). Descriptions are concise (50-150 chars) and explain purpose well. Parameters have types, descriptions, and constraints (enums, min/max). However, output schemas are not documented in the source code, only inferred from API behavior. Error handling is present but generic. No tool annotations (readOnlyHint, etc.) visible in schema definitions.

Tools (4)

fetch_contentread onlyauthsource verified80/100

Fetch and extract clean content through the free TinyFish Fetch API. Each URL is processed independently; per-URL failures appear in errors. This tool does not create browser sessions or run web automation.

get_search_usageread onlyauthsource verified73/100

List Search API usage records for audit and troubleshooting.

list_fetch_usageread onlyauthsource verified73/100

List Fetch API usage records for audit and troubleshooting.

searchread onlyauthsource verified80/100

Search the web through the free TinyFish Search API. Returns ranked titles, snippets, URLs, and optional thumbnail URLs. This tool does not start Agent, Browser, batch, or run lifecycle APIs.

Issues (5)

HIGHtool ↗

Output schemas not documented. Tool descriptions do not specify what fields are returned or their types. LLMs cannot plan downstream tool calls or extract required data without knowing response structure.

searchfetch_contentget_search_usagelist_fetch_usage

MEDIUMtool ↗

Tool annotations missing. Input schemas lack readOnlyHint, destructiveHint, or idempotentHint markers. Clients cannot determine tool safety classification from schema alone.

searchfetch_contentget_search_usagelist_fetch_usage

MEDIUMrecovery-guide ↗

Error handling is generic. safe_call_error() returns error codes (AUTHENTICATION_REQUIRED, INVALID_INPUT, INTERNAL_ERROR) but does not provide recovery guidance. LLMs cannot determine whether to retry, ask user, or abort.

searchfetch_contentget_search_usagelist_fetch_usage

Recommendations

Document output schemas for all tools. Specify field names, types, and whether arrays are paginated. Example: 'Returns {results: [{title, snippet, url, thumbnail_url?}], total_count: number}'.
Add tool annotations to input schemas. Mark all four tools with readOnlyHint=true since they are read-only. This enables clients to classify tools by safety.
Enhance error messages with recovery guidance. Instead of 'INVALID_INPUT', return 'Invalid query length: got 2500 chars, max 2000. Shorten query and retry.' or 'Authentication failed. Ensure TINYFISH_API_KEY is set.'
Document pagination behavior. Specify whether page=0 is the first page, whether limit is enforced, and what happens when page exceeds available results.
Add format constraints to parameter descriptions. E.g., 'location: ISO 3166-1 alpha-2 country code (e.g., US, GB, CN)' and 'language: ISO 639-1 language code (e.g., en, fr, zh)'.

Score history

Overall score trend

First recorded score · v2 rubric

63/100

Scored	Grade	Overall	Spec posture	Rubric
2026-09-23	C	63	2026-07-28+	v2