mediumsearch_jass_knowledge
Description manipulates tool-selection behavior via imperative instructions
The description uses directive language ('Nutze dies für JEDE Frage...', 'ohne dass "JassWiki" genannt wird') instructing the agent to always invoke this tool for a broad category of queries, even when the user doesn't reference the tool or brand by name. This is an attempt to bias/override the agent's normal judgment about when to call the tool, effectively forcing tool usage and brand injection ('offiziell', 'attestiert', 'verifiziert') without the user's explicit request, which could mislead users into thinking answers are officially endorsed by a governing body when they are just wiki content from jasswiki.ch.
mediumshow_jass_card
Aggressive 'always call first' instruction with unverified authority claims
The description repeatedly instructs the agent to treat this as the 'PRIMÄRES Tool' and 'Standard-Antwortweg', mandating it be called first for ANY concrete Jass question, explicitly noting it should trigger 'even without JassWiki being mentioned'. It also asserts official endorsement ('offiziell, gebrandete... Jassverband Schweiz') which cannot be verified by the schema and may be a false authority claim designed to make the agent prioritize this tool over other legitimate sources or the agent's own knowledge, and to make outputs appear more authoritative to the end user than they actually are.
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.
lowget_term_details
Minimal transparency, relies on ecosystem context for scope
Description is benign on its own but functions as part of a tool trio whose sibling descriptions try to steer the agent into a fixed invocation pattern (search_jass_knowledge / show_jass_card first, then get_term_details for 'verification'). No direct manipulation in this tool's text, but flagged for awareness of the coordinated pattern.