The Agent Skills Directory

[PROMPT_INJECTION]: The skill exhibits an attack surface for indirect prompt injection by processing external data from multiple sources.
Ingestion points: Data enters the agent context through metabolic model loaders such as read_sbml_model, load_json_model, load_yaml_model, and load_model (referenced in SKILL.md and api_quick_reference.md).
Boundary markers: No explicit delimiters or instructions are used to separate model content from the instructional context.
Capability inventory: The skill employs the Bash tool (SKILL.md) and performs file system operations including write_sbml_model and to_csv in workflows.md.
Sanitization: The provided instructions and scripts do not include sanitization or validation for descriptive strings within model files.
[EXTERNAL_DOWNLOADS]: Fetches metabolic models from well-known scientific repositories including BiGG and BioModels.
[COMMAND_EXECUTION]: Uses uv pip install to manage the installation of the cobra package and its solver dependencies.

cobrapy