ClearTrust Skills MCP
One MCP endpoint behind the ClearTrust agent skills. Every tool is read-only.
https://skills.cleartrust.cc/mcp
Connect
claude mcp add --transport http cleartrust-skills https://skills.cleartrust.cc/mcp
Any MCP client that speaks Streamable HTTP works. Add Authorization: Bearer cts_… if you have a key.
Skills
- IVT Traffic Forensics ivt-traffic-forensics
Find bots and invalid traffic in web, ad and click logs, with evidence and a blocklist. - MCP Server Security Audit mcp-server-security-audit
Grade an MCP server for tool poisoning, hidden text, shadowing, drift and broken auth. - Ad Creative Malvertising Review ad-creative-malvertising-review
Catch redirects, pop-unders, cloaking and scam creatives before they run. - ads.txt Supply-Chain Check ads-txt-supply-chain-check
Lint ads.txt and app-ads.txt and verify every seller against sellers.json. - Skill Preflight skill-preflight
Run the marketplace security checks on an agent skill before you submit it. - Agent-Readiness Audit agent-readiness-audit
Score how well AI agents and AI search can find, read and use a website. - Headless MCP Service Design headless-mcp-service-design
Design an MCP-first service agents can drive: tool surface, errors, auth, tenancy.
Tools
ivt_analyze | Analyse traffic for invalid traffic (IVT) |
ivt_list_tests | List the IVT tests |
mcp_audit_surface | Audit an MCP tool surface |
mcp_scan_server | Read-only scan of a live MCP server network |
mcp_diff_surfaces | Compare two MCP tool snapshots |
creative_scan | Review an ad creative or tag |
creative_check_landing | Check an ad landing page network |
adstxt_lint | Lint ads.txt text |
adstxt_check_domain | Check a domain's ads.txt against sellers.json network |
skill_preflight | Preflight an agent skill package |
site_agent_readiness | Audit a website for AI-agent readiness network |
Limits and keys
| Free (no key) | 40 calls/day per IP, 8 network calls, 50,000 records per traffic analysis |
| Pro key | 2,000 calls/day, 300 network calls, 400,000 records |
| Team key | 10,000 calls/day, 1,500 network calls |
To get a key, write to [email protected].
Privacy
Inputs are processed in memory and not stored. We keep a daily call count per key or IP address, for limits. Credentials you pass to mcp_scan_server are used for that scan only. Outbound fetches reach public addresses only.