Loading ToolBench
MCP server that exposes conversation histories captured by OpenChat as markdown resources and provides tools to search conversations
Single tool 'search_conversations' has a clear, action-verb name and reasonable description (95 chars). Input schema is present with proper type and description for the 'pattern' parameter. However, the tool lacks critical details: no output schema documentation, no error handling guidance, no parameter constraints (regex validation rules), and no examples of what results look like. The description does not explain when to use this tool vs. browsing resources directly, or what happens if the regex is invalid. Parameter description is minimal (18 chars), below the 72-char baseline for A+ tools. No pagination, no result limits documented. Overall, the tool is functional but underspecified for LLM reasoning.
Search conversation titles by regex pattern. Returns matching resource URIs that can be read individually.
No output schema documented. LLMs cannot infer what fields the response contains or how to chain results to resource reads.
Parameter 'pattern' lacks validation constraints. No guidance on regex syntax, length limits, or what happens if the pattern is invalid (e.g., unclosed bracket, catastrophic backtracking).
No error handling guidance. Tool description does not explain what to do if regex compilation fails, no matches found, or storage is inaccessible.
Parameter description is only 18 characters ('Regex pattern to search conversation titles'). Baseline for A+ tools is 72 chars. Missing context: what constitutes a valid regex? What happens on invalid input?
| Scored | Grade | Overall | Spec posture | Rubric |
|---|---|---|---|---|
| 2026-09-23 | D | 54 | 2026-07-28+ | v2 |
No result limits or pagination documented. If 1000 conversations match, does the tool return all URIs (context explosion) or cap at N results?