Loading ToolBench
Browserless MCP server for Reddit Answers with plain-HTTP challenge solving and token reuse.
Server has 4 tools with complete input/output schemas and descriptions. Tool names follow verb_noun convention (search, follow_up, status, refresh). Descriptions are adequate (80-150 chars) but lack actionable context for LLM selection. Input parameters are typed with Zod and have descriptions. Output schemas are documented. However, parameter descriptions lack constraint details (e.g., UUID format for answerId is declared in schema but not explained in description text). Error handling is absent, no guidance on retryability, user-fixable errors, or recovery paths. No tool annotations (readOnlyHint/destructiveHint) despite clear risk profiles (3 READ_ONLY, 1 WRITE). Missing dependency hints and prerequisites.
Ask a follow-up question in an existing Reddit Answers conversation by reusing a prior answer ID.
Force-refresh the browserless Reddit Answers session and bearer token by replaying Reddit's current JS challenge flow over plain HTTP.
Ask Reddit Answers a brand-new question using a fully browserless bootstrap flow and return the final answer plus follow-up suggestions.
Inspect the current cached Reddit Answers browserless session, token expiry, cookie names, and bootstrap URLs.
No tool annotations (readOnlyHint/destructiveHint/idempotentHint) despite clear risk profiles. reddit_answers_refresh_session is destructive (WRITE) but unmarked; reddit_answers_search and follow_up are read-only but unmarked. LLMs cannot infer safety/retry semantics.
Parameter descriptions lack constraint details. answerId is declared as UUID in schema but description does not explain format or how to obtain it. includeReasoning lacks explanation of what 'reasoning blocks' are or when to enable.
No error handling guidance. Tools lack descriptions of failure modes, retryability, or recovery steps. E.g., what happens if answerId is invalid? Is it retryable? Should the LLM search again?
| Scored | Grade | Overall | Spec posture | Rubric |
|---|---|---|---|---|
| 2026-09-23 | C | 60 | 2025-06-18+ | v2 |
Tool descriptions lack dependency hints and prerequisites. reddit_answers_follow_up requires a prior answerId but does not explain how to obtain one or that reddit_answers_search must be called first.
Output schema fields (quotaRemaining, quotaResetSeconds, patchEventCount, totalEventCount) are undocumented. LLMs cannot infer what these fields mean or when to use them for decision-making.