accept-agent-payments
Pass
Audited by Gen Agent Trust Hub on Sep 17, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONEXTERNAL_DOWNLOADSCOMMAND_EXECUTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill ingests data from external URLs and service responses, creating a surface for potential indirect prompt injection.
- Ingestion points: Content fetched from developers.circle.com and agents.circle.com, and output from circle services inspect and 402 response bodies.
- Boundary markers: Explicit instructions are provided under 'Instruction Priority and Untrusted Data' to treat external metadata as untrusted and to ignore any instructions embedded within it.
- Capability inventory: The skill possesses Edit, Write, and restricted Bash capabilities, but excludes the circle services pay command to mitigate financial risk.
- Sanitization: The skill contains strict validation requirements for all command arguments, checking for shell metacharacters and ensuring format compliance for URLs and addresses.
- [EXTERNAL_DOWNLOADS]: The skill uses curl to retrieve documentation and service listings from developers.circle.com and agents.circle.com, which are official Circle resources.
- [COMMAND_EXECUTION]: The skill uses a restricted set of shell commands for system discovery and package metadata retrieval. It provides a safety framework for constructing shell commands to prevent injection from user or service-supplied data.
Audit Metadata