The Agent Skills Directory

[COMMAND_EXECUTION]: The script scripts/repomix_batch.py executes the repomix CLI and npx using subprocess.run. The implementation correctly uses a list of arguments instead of a raw shell string, effectively preventing shell injection vulnerabilities.
[EXTERNAL_DOWNLOADS]: The skill supports processing remote repositories via npx repomix --remote. This is a primary feature of the tool and operates on user-provided URLs or repository identifiers.
[PROMPT_INJECTION]: The skill inherently creates an indirect prompt injection surface because it is designed to ingest external repository content for AI analysis. Instructions embedded in a processed codebase (e.g., in a README or code comment) could potentially influence the behavior of the AI model that later processes the generated output. This is a known risk factor for any tool that aggregates untrusted data for LLM consumption.
[DATA_EXFILTRATION]: The EnvLoader class in scripts/repomix_batch.py is designed to load environment variables from multiple .env file locations. This is used to manage configuration and secrets (like GitHub tokens) for the underlying repomix tool. This is standard practice for local agent utilities.

repomix