The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill instructs the agent to analyze and modify user-provided files (such as pyproject.toml, requirements.txt, and Dockerfile) to optimize container builds. Maliciously crafted content in these untrusted source files could potentially attempt to influence the agent's actions during the modification process.
Ingestion points: Processes user project configuration and Dockerfiles via file-reading tools like Read and Grep.
Boundary markers: No specific delimiters or "ignore" instructions are provided for handling untrusted data in the prompt templates.
Capability inventory: The agent has access to shell execution (Bash), file modification (Edit, Write), and web access (WebFetch).
Sanitization: No explicit sanitization or validation of external data is defined in the instructions.
[COMMAND_EXECUTION]: The skill includes suggested shell commands for the agent to build images, verify imports, and perform security audits using standard tools like docker and pip-audit. These are standard development operations and do not involve untrusted remote execution.

python-containers