sales-commission-tier-check-free
Pass
Audited by Gen Agent Trust Hub on Sep 16, 2026
Risk Level: SAFEPROMPT_INJECTIONINDIRECT_PROMPT_INJECTIONMETADATA_POISONING
Full Analysis
- [PROMPT_INJECTION]: The SKILL.md file contains a pre-formatted instructional block designed for users to copy and paste into their conversation with the AI agent. This block instructs the agent to perform a multi-step process: installing an external payment tool ('npx -y @alipay/agent-payment@latest install'), purchasing a specific skill ID, and using 'pre-authorization' language ('I have authorized... do not confirm for me') to automate the transaction and reduce user oversight during the installation process.\n- [INDIRECT_PROMPT_INJECTION]: The skill processes user-provided data, such as sales commission tables, through scripts/run.mjs and its engine. Although the script currently treats this input as data for mathematical validation, there is a risk for indirect prompt injection as documented in the following chain:\n
- Ingestion points: Input is ingested via the 'text' field in scripts/run.mjs.\n
- Boundary markers: None present. The skill does not use delimiters or warnings to separate data from potential instructions.\n
- Capability inventory: The skill has access to Bash and Read tools, and uses Node.js file system modules.\n
- Sanitization: No sanitization is performed against natural language instructions embedded within the cells of the processed tables.\n- [METADATA_POISONING]: The skill's README documentation (SKILL.md) is used to store and promote pre-written agent instructions that guide the agent toward purchasing and installing secondary skills using automated authorization language, which could influence the agent's behavior beyond the scope of the primary skill.
Audit Metadata