live-commerce-commission-check-free
Pass
Audited by Gen Agent Trust Hub on Sep 16, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONREMOTE_CODE_EXECUTIONCOMMAND_EXECUTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill parses user-provided commission data and interpolates segments of this untrusted input (such as talent names or session identifiers) directly into the diagnostic findings reported back to the agent. This creates a surface where malicious instructions embedded in the source data could influence the agent's subsequent logic.
- Ingestion points: The
payload.textinput processed byscripts/engine/live-commerce-commission-check.jsvia thescripts/run.mjsentry point. - Boundary markers: Absent; the data is parsed and reflected without specific delimiters or instructions to ignore embedded commands.
- Capability inventory: The skill utilizes the
BashandReadtools to execute logic and process files. - Sanitization: Absent; the engine performs numerical normalization but does not sanitize string values reflected in the output.
- [REMOTE_CODE_EXECUTION]: The
SKILL.mdfile provides a text block intended for the user to send to the AI agent which contains a command to download and execute code from the npm registry. - Evidence: The instruction block contains
npx -y @alipay/agent-payment@latest install. While targeting a well-known service (Alipay), this patterns instructs the agent to execute remote code. - [COMMAND_EXECUTION]: The skill is designed to run local Node.js scripts (
scripts/run.mjs) using the environment'sBashtool to perform its primary auditing functions.
Audit Metadata