The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The installation instructions require cloning an external code repository from a third-party GitHub account (tvytlx/ai-agent-deep-dive) that does not match the identified vendor.
[REMOTE_CODE_EXECUTION]: The troubleshooting section suggests installing the Poetry dependency manager by piping a remote script from https://install.python-poetry.org directly to the Python interpreter.
[INDIRECT_PROMPT_INJECTION]: The framework implements an agent loop that ingests untrusted user input to trigger various skills, creating a potential vector for indirect instructions.
Ingestion points: User input passed via the CLI query argument.
Boundary markers: The framework lacks explicit delimiters or instructions to prevent the agent from obeying instructions embedded within user queries.
Capability inventory: The system is designed to load and execute Python modules as skills from a local directory.
Sanitization: The provided examples do not include logic for sanitizing or validating user-supplied queries before they are processed by the agent core.
[CREDENTIALS_UNSAFE]: The documentation and code examples demonstrate the usage of OPENAI_API_KEY and suggest storing secrets in .env files, which is a standard but sensitive configuration pattern.

ai-agent-deep-dive-teaching-framework