The Agent Skills Directory

[COMMAND_EXECUTION]: The skill relies on Bash(curl:*) and Bash(node:*) to perform system diagnostics. It defines procedures in TypeScript and Bash that execute local commands such as curl, openssl, and dig to verify API availability, TLS certificate status, and DNS resolution.- [DATA_EXFILTRATION]: The skill accesses sensitive credentials from environment variables (ABRIDGE_CLIENT_SECRET, ABRIDGE_ORG_ID). It implements redaction logic in both the TypeScript code and the Bash script to ensure that these secrets are not printed to the console or saved into the debug bundle file. Network operations are performed to check the health status of endpoints configured via user-provided environment variables.- [PROMPT_INJECTION]: The skill contains an indirect prompt injection surface as it processes untrusted data from an external API into the agent's context.
Ingestion points: Data is retrieved from the /encounters/sessions endpoint within src/debug/abridge-debug-bundle.ts.
Boundary markers: The skill does not employ specific delimiters or instruction-ignore markers when interpolating external API data into the diagnostic output.
Capability inventory: The skill possesses Write capabilities for local file creation and Bash for shell execution, which could be targeted by instructions embedded in external data.
Sanitization: A redactPhi function is used to filter sensitive healthcare identifiers using regular expressions, providing a layer of data privacy but not addressing instructional injection risk.

abridge-debug-bundle