Loading ToolBench
Fetches latest GitHub commits, generates AI summaries, and posts to Bluesky. Provides MCP tools for automated social media posting from code repositories.
Both tools have identical, minimal definitions with critical gaps. Tool names lack action verbs (createBskyPost, createBskyPostFromGithub are vague about what 'create' means in context). Descriptions are under 100 chars and lack WHEN/WHY guidance. Input schemas present but parameters ('day', 'key') are poorly named and undescribed. No output schema documented. Parameters expose 'key' as a credential parameter, violating secret-injection pattern. No error handling guidance. Tools appear duplicated across two server implementations with no clear distinction.
Fetch latest GitHub commit, summarize it, and post to Bluesky.
Fetches the latest GitHub commit, generates an AI summary, and posts it to Bluesky.
Credentials exposed as tool parameter: 'key' parameter violates secret-injection pattern. API keys must never appear in tool parameters, they leak into logs and traces. Use server-side environment variable injection instead.
Parameter 'day' is ambiguous and undescribed. Does it mean a date string, day-of-week, or a numeric day ID? No type hint, format constraint, or description provided. Parameter 'key' lacks any description.
Tool names lack action clarity. 'createBskyPost' does not convey that it fetches GitHub commits, summarizes them, and posts to Bluesky. Name should reflect the full intent or be split into separate tools (fetch_commit, summarize_commit, post_to_bluesky).
| Scored | Grade | Overall | Spec posture | Rubric |
|---|---|---|---|---|
| 2026-09-23 | F | 37 | 2026-07-28+ | v2 |
No output schema documented. LLMs cannot plan downstream actions or extract return values. What does the tool return? Success message? Post URL? Commit hash? This must be explicitly documented.
Duplicate tools across implementations (CLI_version and Claude_version) with identical signatures but no clear distinction. Why are there two? This confuses tool discovery and suggests incomplete refactoring.