The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill ingests and processes data from remote HTTPS endpoints via the execute-endpoint and probe-endpoint tools.
Ingestion points: Data is received from arbitrary HTTPS URLs passed as arguments to the CLI in x402.ts.
Boundary markers: The AGENT.md does not specify delimiters or instructions for the agent to ignore potential instructions embedded within the tool output.
Capability inventory: The skill can perform network writes (POST/PUT) and initiate blockchain transactions involving wallet funds.
Sanitization: No explicit sanitization or schema validation of the remote API response is performed before the content is returned to the agent context.
[DYNAMIC_EXECUTION]: The scaffold-endpoint and scaffold-ai-endpoint commands generate new TypeScript projects and configuration files on the local file system.
The tool writes Hono.js application code and Cloudflare Worker configurations to user-defined directories based on endpoint configurations supplied to the command.
While intended for project creation, this capability allows the agent to generate new executable code structures that incorporate potentially untrusted input, requiring user verification before deployment.

x402