Loading ToolBench
A Model Context Protocol (MCP) server for Reddit with full read AND write operations - create posts, reply, edit, and delete content.
Reddit MCP server has 11 tools with consistent naming (verb_noun pattern) and descriptions present for all tools and parameters. However, descriptions are generic and lack LLM-optimization guidance. Input schemas are properly typed with JSON Schema, but output schemas are not documented. Error handling uses UserError but lacks recovery guidance. No tool annotations (readOnlyHint/destructiveHint) despite clear risk stratification (9 READ_ONLY, 2 WRITE). Parameter descriptions lack format constraints, ranges, and dependency documentation. Missing pagination support on list-returning tools (getTopPosts, searchReddit, getUserPosts, getUserComments). No idempotency or confirmation patterns for destructive operations (createPost, replyToPost).
Create a new post in a subreddit with title and content
Fetch comments for a specific Reddit post with sorting and limit options
Fetch detailed information about a specific Reddit post including title, content, stats, and engagement analysis
Fetch detailed information about a subreddit including description, stats, and community analysis
Fetch top posts from a subreddit within a specified time period
Fetch a list of currently trending subreddits
Fetch comments by a specific Reddit user with sorting and time filter options
Fetch detailed information about a Reddit user including karma, account status, and activity analysis
No tool annotations (readOnlyHint/destructiveHint/idempotentHint) despite clear risk stratification. createPost and replyToPost are WRITE operations but lack destructiveHint annotation to signal irreversibility to agents.
Output schemas not documented. Tools return formatted text responses but LLMs cannot parse structure for downstream chaining. No documented return types for any tool.
Parameter descriptions lack format constraints and ranges. 'limit' parameters (default 10-100) have no min/max bounds documented. 'sort' and 'time_filter' parameters accept free-form strings with no enum constraints or valid value documentation.
| Scored | Grade | Overall | Spec posture | Rubric |
|---|---|---|---|---|
| 2026-09-23 | D | 50 | 2026-07-28+ | v2 |
Fetch posts by a specific Reddit user with sorting and time filter options
Reply to a Reddit post with a comment
Search Reddit for posts matching a query with optional subreddit, sort, and time filters
No pagination support on list-returning tools. getTopPosts, searchReddit, getUserPosts, getUserComments return fixed limits (10-100) with no offset/cursor or total_count. Large result sets will blow context window.
Destructive operations (createPost, replyToPost) lack confirmation or dry-run patterns. No recovery guidance in error messages. Agents cannot safely preview or confirm before executing irreversible actions.