Loading ToolBench
MCP server for Alderpost Intelligence API — 8 endpoints bundling premium data sources (VirusTotal, People Data Labs, Hunter.io, AbuseIPDB, Qualys SSL Labs, NIH RxNorm, US Census, ESPN, The Odds API). Domain security, company intel, threat analysis, compliance, sales intelligence, sports, property, and health data. Pay-per-call via x402 USDC on Base.
Alderpost MCP has 8 tools with clear verb_noun naming (domain_shield, company_xray, threat_pulse, etc.) and detailed descriptions averaging 156 characters. However, critical gaps undermine schema and composition quality: (1) Input schemas use Zod but lack explicit type annotations beyond z.string(), no enums, ranges, or format constraints visible; (2) All parameters lack minimum/maximum bounds or validation rules in descriptions; (3) No output schemas are documented, tool responses are opaque JSON from external APIs with no guidance on fields, structure, or pagination; (4) Parameter descriptions mention 'e.g.' values (stripe.com, hubspot.com, 8.8.8.8) which LLMs may reuse literally; (5) No error handling guidance, callEndpoint silently returns raw API responses without categorization or recovery hints; (6) No tool composition support, responses are unstructured API dumps that downstream tools cannot chain from; (7) Tool descriptions reference pricing ($0.10-$0.15) but do not explain how results should guide agent decisions. The server meets baseline naming and description length standards (18-27 char names, 150-200 char descriptions) but fails on the precision required for production LLM agents.
Company intelligence powered by People Data Labs + Hunter.io. Industry, employee count, revenue estimate, tech stack, verified email contacts, social presence. 9 data sources scored 0-100. Price: $0.15 USDC on Base.
IT security compliance audit with Qualys SSL Labs grading. Email auth, SSL/TLS, OWASP headers, cookies, privacy, DNSSEC, hosting, data exposure. 8 checks letter graded. Price: $0.15 USDC on Base.
Domain security scan with VirusTotal malware detection (70+ antivirus engines). SPF, DKIM, DMARC, SSL, MX, DNSSEC, domain info. 8 checks scored 0-100 with prioritized recommendations. Price: $0.12 USDC on Base.
Health intelligence with NIH RxNorm drug interaction checking + FDA data. Drug labels, adverse events, recalls, nutrition, drug-to-drug interactions with severity ratings. 7 sources risk scored 0-100. Price: $0.10 USDC on Base.
Location intelligence with US Census Bureau + OpenWeather data. Geocoding, nearby amenities, schools, elevation, demographics, walkability. 7 sources scored 0-100. Price: $0.10 USDC on Base.
No output schemas documented. All 8 tools return raw JSON via callEndpoint(). LLMs cannot know what fields to expect, making downstream tool composition impossible and forcing agents to guess at response structure. This violates the pattern:response-shaper baseline where 100% of A+ tools have documented return types.
Input parameters lack validation constraints. domain (6 tools), target, address, query, sport, team are all z.string() with no enums, length limits, format patterns, or ranges. Descriptions mention 'e.g. stripe.com' which LLMs may treat as the only valid input. This violates pattern:constrained-input and review:param-validation-rules.
| Scored | Grade | Overall | Spec posture | Rubric |
|---|---|---|---|---|
| 2026-09-22 | D | 54 | 2026-07-28+ | v2 |
Sales intelligence powered by People Data Labs + Hunter.io. Web presence, tech stack, verified contacts, social signals, contact readiness. 7 sources scored 0-100. Price: $0.12 USDC on Base.
Pre-game sports intelligence with ESPN + The Odds API + Claude AI synthesis. Live standings, odds comparison, AI-generated game analysis. Price: $0.12 USDC on Base.
Threat intelligence with VirusTotal (70+ engines) + AbuseIPDB (30K+ community reporters). Blacklists, reverse DNS, open ports, SSL analysis, email security. 7 sources scored 0-100. Price: $0.10 USDC on Base.
No error handling or recovery guidance. callEndpoint() returns status + raw data with no categorization of errors as retryable, user-fixable, or fatal. HTTP errors (4xx, 5xx) are not distinguished from successful responses. Agents have no actionable error responses. Violates pattern:recovery-guide and pattern:error-classification.
Example values in descriptions (stripe.com, hubspot.com, 8.8.8.8, lakers) are treated by LLMs as literal valid inputs rather than examples. Should use enum constraints or format patterns instead. Violates review:param-validation-rules.
No tool composition support. Responses are opaque API dumps with no promise of stable fields, IDs, or references for chaining. If an agent wanted to sequence domain_shield → compliance_check → threat_pulse, it has no guarantee the output from one matches the input format of the next. Violates pattern:tool-chain.
sports_edge accepts optional 'team' parameter but does not document what sport keys are valid (description says 'nba, nfl, mlb, nhl, mls, epl' but as free-form text, not enum). LLMs may invent invalid keys. team parameter has no guidance on format (case sensitivity, spelling variants).
health_signal 'type' parameter accepts 'drug, food, or auto' but lacks enum constraint. LLMs may pass 'medicine', 'supplement', 'food item', or other variants, causing silent API failures or unexpected behavior.
No pagination support documented or visible. Tools that might return large result sets (company_xray with 9 sources, threat_pulse with 7 sources, compliance_check with 8 checks) have no limit, offset, or page parameters. Result sets could bloat context window. Violates pattern:paginated-result.