01Safety
0.0 / 30
What changed in the harness
Selection accuracy 95→95, token cost up 2%, unconfirmed writes 0%→0%.
Category breakdown
Where the score comes from.
Earned points across the four signals Gradable measures. Safety and Legibility are scored out of 30; Economics and Discoverability are scored out of 20.
02Legibility
26.8 / 30
03Economics
20.0 / 20
04Discoverability
17.5 / 20
Highest-impact fix
Estimated gain +30 pointsAdd explicit identity and permission preflight tools
Expose machine-readable principal/tenant confirmation and a non-mutating permission check so agents can verify both before destructive actions.
Description evidence
Defects and rewrites.
8 defects found across the exposed tool descriptions. Suggested rewrites make purpose, inputs, boundaries, and returns easier for an agent to understand.
| Tool | Defect types | Suggested rewrite |
|---|---|---|
domain_shield |
no_return_description |
Domain security scan with VirusTotal malware detection (70+ antivirus engines). SPF, DKIM, DMARC, SSL, MX, DNSSEC, domain info. Returns a report of 8 checks scored 0-100 with prioritized recommendations. Price: $0.12 USDC on Base. |
company_xray |
no_return_description |
Company intelligence powered by People Data Labs + Hunter.io. Returns a scored profile covering industry, employee count, revenue estimate, tech stack, verified email contacts, and social presence across 9 data sources scored 0-100. Price: $0.15 USDC on Base. |
threat_pulse |
no_return_description |
Threat intelligence with VirusTotal (70+ engines) + AbuseIPDB (30K+ community reporters). Returns a scored threat report covering blacklists, reverse DNS, open ports, SSL analysis, and email security across 7 sources scored 0-100. Price: $0.10 USDC on Base. |
compliance_check |
no_return_description |
IT security compliance audit with Qualys SSL Labs grading. Returns letter grades for 8 checks covering email auth, SSL/TLS, OWASP headers, cookies, privacy, DNSSEC, hosting, and data exposure. Price: $0.15 USDC on Base. |
prospect_iq |
no_return_description |
Sales intelligence powered by People Data Labs + Hunter.io. Returns a scored prospect profile covering web presence, tech stack, verified contacts, social signals, and contact readiness across 7 sources scored 0-100. Price: $0.12 USDC on Base. |
sports_edge |
no_return_description |
Pre-game sports intelligence with ESPN + The Odds API + Claude AI synthesis. Returns live standings, odds comparisons from 15+ bookmakers, and AI-generated game analysis. Price: $0.12 USDC on Base. |
property_intel |
no_return_description |
Location intelligence with US Census Bureau + OpenWeather data. Returns a scored report covering geocoding, nearby amenities, schools, elevation, demographics, and walkability across 7 sources scored 0-100. Price: $0.10 USDC on Base. |
health_signal |
no_return_description |
Health intelligence with NIH RxNorm drug interaction checking + FDA data. Returns a risk-scored report covering drug labels, adverse events, recalls, nutrition, and drug-to-drug interactions with severity ratings across 7 sources scored 0-100. Price: $0.10 USDC on Base. |
Selection evidence
Confusable tool pairs.
1 pair where similar names or overlapping descriptions may send an agent toward the wrong tool.
| Tool A | Tool B | Confidence | Why they collide |
|---|---|---|---|
company_xray |
prospect_iq |
medium | Both are near-identical intelligence tools (same providers, same domain input, both rate sources 0-100) and overlap on tech stack and verified contacts. A task like 'get an intelligence report on this company's tech stack and contacts' maps to both; only firmographics (employees/revenue) vs. sales signals (contact readiness) separates them, so an agent may pick the wrong one. |
Compare the field