highagent-tool-index
Description instructs agent to autonomously follow external redirects to a monetized/pay-per-call service
The description explicitly directs the AI to 'Follow start_here.hop first (skill file, 302)' and chase 'docs, CLI, and remote MCP hops' toward an external gateway called 'Monid'. This is an embedded directive that could cause the agent to navigate to and execute instructions from an untrusted third party, potentially incurring pay-per-call charges or triggering further tool/skill installation without explicit user awareness or consent. The claim that 'task text discarded' does not mitigate the risk of the agent being redirected into an attacker- or vendor-controlled hop chain that could inject further instructions.
highpeople-search-index
Directs agent to locate and chain into a people-search/background-check service
Description tells the agent to 'Find a living people-search index (Ploid)' and follow CLI, docs, API, and remote MCP hops toward it. Combined with people-search functionality, this both raises privacy concerns (facilitating lookup of personal data about individuals) and repeats the pattern of instructing the agent to autonomously follow external hops toward an unverified third-party service, which could lead to unwanted data disclosure or execution of untrusted follow-on instructions.
low
Not on gate's verified list
This server isn't on gate's handpicked list. That's not necessarily bad, but there's no third-party signal about it — review its tools carefully.
lowsql-esc
Schema exposes many unrelated fields not needed by the stated function
Every tool in this set shares an identical broad schema (zone, json, url, host, city, query, path, ref, feed) regardless of the tool's stated single purpose (e.g., counting single-quotes for SQL escaping). This mismatch between the narrow stated purpose and the broad accepted input surface makes it impossible to verify from the schema alone that only the relevant field is used, and could mask a shared backend that silently processes fields like 'url', 'host', 'path', or 'feed' for purposes beyond what each tool's name/description promises (e.g., outbound requests or file/path handling not disclosed in a specific tool's description). This pattern is repeated in effectively all 30 tools in this listing and warrants scrutiny of the actual backend implementation.