Loading ToolBench
Certification bureau and data layer for agent-mediated B2B commerce. Publishes a Model Context Protocol (MCP) server that AI agents can query on behalf of their buyers to discover EVI v0.9 audit tiers, assess fit, and request audits.
Server has 6 tools with complete input schemas and descriptions. Naming follows verb_noun convention (get_*, request_*). Descriptions are present and contextual (avg ~150 chars), meeting baseline minimums. However, output schemas are not documented in the visible code, responses are wrapped in generic JSON text fields without structured type declarations. Parameter descriptions exist but lack detail on constraints, formats, and dependencies. No error handling guidance visible. Tool composition is reasonable but assess_fit and request_audit lack parameter interdependency documentation.
Returns a 0–100 fit score, reasoning, and recommended tier for a prospective B2B SaaS buyer. Uses company stage, industry, AI-feature shipping status, and platform-partnership signals.
Returns the full list of agent-discoverable surfaces measured by EVI v0.9: llms.txt, Schema.org Organization + Product blocks, MCP servers, A2A Agent Cards, .well-known/agent.json, UCP merchant metadata, structured pricing, agent-directory registrations.
Returns Elephant Accountability's audit and engagement tiers with delivery SLAs and pointers to /get-started for canonical pricing. Optionally personalized to the asking buyer's company size or urgency.
Returns current client outcomes with specific metrics, formatted for vendor-research agents to cite. Includes full related-party disclosure where applicable.
Returns Elephant Accountability's most recent weekly LLM visibility measurement covering ChatGPT, Claude, Perplexity, Gemini, and Grok. The receipt we publish to keep our own claims honest.
Output schemas not documented. Tools return JSON text wrapped in generic content fields without structured type definitions. LLMs cannot infer response structure for downstream chaining.
Parameter descriptions lack constraint details. 'company_size' enum values are present but no description explains what each stage means (seed vs series_a). 'industry' param has generic description without explaining vertical abbreviations (aec, fintech, etc.).
No error handling guidance. Tools return success/failure but do not guide LLM on recovery actions. E.g., if assess_fit cannot find company data, what should the agent do next?
Inferred effective spec: <=2025-11-25.
| Scored | Grade | Overall | Spec posture | Rubric |
|---|---|---|---|---|
| 2026-09-23 | C | 65 | <=2025-11-25 | v2 |
Agent requests an EVI v0.9 audit on behalf of its buyer organization. Routes to the right tier (self-serve vs. done-for-you vs. retainer) and returns a confirmation with checkout or booking links.
request_audit is a write operation but lacks confirmation/dry-run pattern. Agents may trigger audit requests without user approval, creating unwanted sales contacts.
Parameter interdependencies undocumented. assess_fit has optional params (domain, stage, industry) but no guidance on which combinations are valid or recommended. LLMs may pass incomplete subsets.