The Agent Skills Directory

[SAFE]: The skill serves as a reference guide for the Botpress ADK testing framework. All code examples and documentation follow legitimate software development and testing practices.
[REMOTE_CODE_EXECUTION]: The skill mentions and demonstrates the use of the @botpress/evals package and the adk CLI. These are identified as official vendor resources from 'botpress' used for local development and testing purposes. There are no patterns of executing untrusted remote code or scripts from unknown sources.
[DATA_EXFILTRATION]: No evidence of data exfiltration was found. Network references are limited to localhost:3001, which is the standard port for the local developer console, and placeholders in example code.
[CREDENTIALS_UNSAFE]: No hardcoded credentials or sensitive file paths were detected. All identifiers used in examples (e.g., test-user-1, ORD-123) are generic placeholders.
[OBFUSCATION]: The content was analyzed for hidden characters, Base64 encoding, and homoglyph attacks; no suspicious obfuscation techniques were found.
[INDIRECT_PROMPT_INJECTION]: While the skill describes processing user messages and events in tests, it provides structured patterns and safety-conscious testing methodologies (e.g., negative testing patterns). It does not introduce runtime vulnerabilities to indirect injection.

adk-evals