solana-clawd-agentic-commerce
Pass
Audited by Gen Agent Trust Hub on Jul 9, 2026
Risk Level: SAFECOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
- [PROMPT_INJECTION]: The skill interacts with external providers and endpoints for paid services, creating a surface for indirect prompt injection.
- Ingestion points: Data and responses from external service providers and API endpoints during the planning and execution phases (SKILL.md).
- Boundary markers: The skill instructions require the agent to load 'pay-policy.ts' and use
PayAutonomyClient.planto validate calls against policy, acting as a functional boundary. - Capability inventory: The skill can execute shell commands via the
payCLI andnodescripts (SKILL.md). - Sanitization: Instructions mandate refusing calls that hit denied URL patterns or target non-allowlisted endpoints.
- [COMMAND_EXECUTION]: The skill executes shell commands to initialize commerce servers and run example scripts using local framework files.
- Evidence: Execution of
pay --sandbox server start openclawd-framework/pay/solana-clawd-agent-commerce.yml --debuggerandnode --import tsx/esm examples/agent-commerce.ts.
Audit Metadata