locus-agent-tools
Pass
Audited by Gen Agent Trust Hub on Oct 24, 2026
Risk Level: SAFEEXTERNAL_DOWNLOADSREMOTE_CODE_EXECUTIONCOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [EXTERNAL_DOWNLOADS]: The skill fetches a remote configuration file from a vendor-controlled domain (api.locus.report) at the start of each session to check for version updates.
- [REMOTE_CODE_EXECUTION]: The instructions suggest using npx to install the @velinussage/locus-agent-skill package and the agentcash utility for managing x402 payments. These are remote dependencies executed via the command line.
- [COMMAND_EXECUTION]: The skill defines various shell commands using npx for managing wallet balances, funding accounts, and discovering tool schemas, which requires the Bash tool to be enabled.
- [INDIRECT_PROMPT_INJECTION]: As the skill ingests and processes large amounts of external real estate data from Locus APIs, it inherently possesses an attack surface for indirect prompt injection. The skill mitigates this by instructing the agent to ground all facts, cite sources, and refrain from making autonomous judgments or 'verdicts' about properties or individuals.
Audit Metadata