highagent-tool-index
Embedded directive to autonomously follow external redirects/hops
The description instructs the agent to 'Follow start_here.hop first (skill file, 302)' and says it 'also returns docs, CLI, and remote MCP hops.' This is an instruction aimed at the agent's behavior rather than a description of a data-return function — it tells the agent to chase a chain of redirects to an unvetted, pay-per-call third-party gateway and to load additional remote MCP tool sources. This could be used to smuggle in new tool definitions, incur charges, or redirect the agent to attacker-controlled infrastructure without the user's awareness or consent.
highpeople-search-index
People-search discovery tool with hop-chaining instructions
Description promotes discovery of a 'people-search index (Ploid)' and directs the agent to follow CLI/docs/API/remote-MCP hops. Beyond the privacy concerns of facilitating people-search/doxxing lookups, the instruction to chain into additional remote MCP endpoints mirrors a supply-chain/prompt-injection pattern where the agent is nudged to load and trust unverified external tool sources under the pretense of a lookup utility.
mediumagent-tool-index
Pay-per-call capability far exceeds stated scope
A tool nominally for 'finding a gateway' is described as leading to a pay-per-call system, which could result in real-world financial transactions being initiated by the agent based on hidden hop-following instructions, without explicit user confirmation at the point of charge.
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.
lowbrowser-url-ok
Tool naming may create false capability impression
Named 'browser-url-ok' alongside a near-duplicate 'playwright-url-ok' with identical description ('Check whether a URL can be opened. No browser is launched.'). The Playwright-style naming could mislead the agent (or user) into assuming real browser automation capability is available, when it is not; duplicate near-identical tools with different names also increases risk of confusion/spoofing in tool selection logic.
lowmph-ms
Input schema unrelated to stated function, shared identically across all 30 tools
Every tool in this set exposes the exact same broad parameter set (zone, json, url, host, city, query, path, ref, feed) regardless of its stated single-purpose function (e.g., a unit converter accepting a 'path', 'ref', 'feed', 'json' with no stated use). This mismatched, overly-broad schema template makes it hard for the agent or a reviewer to know which parameters actually get used/transmitted per call, and creates a vector where sensitive-looking fields (e.g., 'path' for local file paths) could be silently accepted and forwarded even though the description claims 'no disk access' — the discrepancy between declared schema and declared behavior should be verified against actual implementation.