www.cloudworldmodel.ai

Very long descriptions are sometimes used to bury instructions to the AI. Skim it before allowing.

Very long descriptions are sometimes used to bury instructions to the AI. Skim it before allowing.

Very long descriptions are sometimes used to bury instructions to the AI. Skim it before allowing.

Very long descriptions are sometimes used to bury instructions to the AI. Skim it before allowing.

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.

The description contains the instruction: 'Do not call api.spec to learn the simulation workflow — the tool descriptions in this session contain everything needed.' This is an instruction aimed at steering the agent's behavior (telling it not to consult another reference/tool) rather than describing the tool's own function. While likely benign here, this pattern—embedding directives in a tool description that discourage the agent from verifying information elsewhere—is a technique that could be misused in other contexts to prevent an agent from cross-checking suspicious behavior. Worth noting even though no secret-harvesting or exfiltration is present.

Multiple tools (inject_failure, recover_resource, delete, create, step, metrics) repeat 'Authenticate with an API key to unlock all 62 tools...'. This is not malicious but is an unusual amount of repeated marketing-style language embedded in functional tool descriptions, which could condition the agent to seek/request API key authentication more readily than warranted. No actual credential request or exfiltration occurs, so this is informational only.