The Agent Skills Directory

[PERSISTENCE]: Documentation and setup guides (README.md, QUICK_REFERENCE.md, SKILL.md) instruct users to append environment variable export commands to shell profile files such as ~/.bashrc or ~/.zshrc. This is a persistence mechanism that modifies system startup scripts to maintain credentials across sessions.
[INDIRECT_PROMPT_INJECTION]: The skill ingests untrusted user input in the form of diagram descriptions and interpolates it directly into prompts for both the image generation and quality review models in scripts/generate_schematic_ai.py.
Ingestion points: User-provided diagram descriptions entered via the CLI are used in the generate_iterative method.
Boundary markers: The prompt templates lack delimiters or specific instructions to the model to ignore embedded commands within the user input.
Capability inventory: The skill has capabilities to write files (images and JSON logs) and perform network operations (API calls to OpenRouter).
Sanitization: No validation, escaping, or filtering is performed on user-supplied strings before prompt interpolation.
[COMMAND_EXECUTION]: The wrapper script scripts/generate_schematic.py uses subprocess.run() to execute the internal generation script. This pattern facilitates the execution of shell commands.

scientific-schematics