Loading ToolBench
Track how your brand appears in Perplexity, ChatGPT, Claude and Gemini search results directly inside Claude, Cursor, Codex, Windsurf and any MCP-compatible AI agent. Measure GEO visibility, citations, and share-of-voice across AI search platforms.
Two tools with comprehensive descriptions (200+ chars each) and well-structured input schemas using Zod validation. Both tools have clear verb-noun naming (audit_*, check_*) and detailed parameter documentation. However, output schemas are not explicitly documented in the visible code, only inferred from Apify actor responses. Error handling exists but lacks actionable recovery guidance. Tool descriptions are excellent for LLM selection but could benefit from explicit output field documentation.
Run a comprehensive AI-search brand visibility audit across Perplexity, ChatGPT (with live web search), Claude (with native web search) and Gemini (with Google Search grounding) using standard GEO query templates. Returns per-check records with mention rate, position score, sentiment, citation detection, and competitor co-mentions, plus a run summary with an overall GEO score (0–100) and actionable recommendations. Use for baseline audits, weekly monitoring runs, or before/after content-push measurement. Do NOT use when the user already has specific custom questions — use `check_brand_in_queries` for that. Backed by a production Apify actor with native web search on all four platforms. Pay-per-result: $0.08 per brand × query × platform check.
Submit 1–10 custom questions to Perplexity, ChatGPT, Claude, and Gemini (all with live web search) and analyze each response for brand mentions, competitor co-mentions, citation detection, sentiment, and mention position. Use when the user has specific prompts they want to test — e.g. 'What tool does ChatGPT recommend for backlink research?' or 'Which CRM does Perplexity cite for small business?'. Do NOT use for comprehensive template-driven audits — use `audit_brand_visibility` for that. Returns one record per query × platform with full mention analysis. Pay-per-result: $0.08 per brand × query × platform check.
Output schema not documented in tool metadata. Apify response structure (BrandCheckRecord, RunSummary) is defined in types.ts but not exposed in tool descriptions or input schema. LLMs cannot plan downstream operations without knowing what fields to expect.
Error responses lack actionable recovery guidance. When Apify actor fails (status !== SUCCEEDED), error message includes run URL but does not suggest next steps (retry, check token, reduce query count). LLMs need explicit guidance on whether to retry or escalate.
No dry-run or confirmation step for cost-incurring operations. Both tools trigger billable Apify actor runs ($0.08 per check). No mechanism to preview cost or confirm before execution. Agents may accidentally trigger expensive audits.
| Scored | Grade | Overall | Spec posture | Rubric |
|---|---|---|---|---|
| 2026-09-23 | B | 71 | 2025-06-18+ | v2 |
STDIO transport only. Server is not remotely accessible and cannot be used by hosted MCP clients (Claude web, Cursor Cloud, etc.). Limits deployment to local/desktop environments.
Missing pagination/result limits documentation. Tools can return up to 500 items (dataset.listItems limit: 500) but descriptions do not mention this cap or how to handle large result sets. No offset/limit parameters for controlling result volume.