The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill fetches and installs the mfapy library directly from a specific commit on a GitHub repository (github.com/fumiomatsuda/mfapy.git). This source is external to standard package registries and does not belong to the identified trusted vendor list.
[DYNAMIC_EXECUTION]: The mfapy engine generates and executes numerical functions at runtime based on the provided metabolic network model. The skill implements a mitigation strategy by validating all identifiers using a restricted regular expression ([A-Za-z][A-Za-z0-9]{0,39}) before they are passed to the internal generator, reducing the risk of code injection.
[INDIRECT_PROMPT_INJECTION]: The skill processes complex scientific data provided in JSON format, creating a surface for potential indirect injection via malicious model definitions or measurement data.
Ingestion points: Inputs are ingested through the read_json function in scripts/_mfa_model.py which loads data from files into the agent's context.
Boundary markers: The skill enforces a strictly defined JSON schema (version 1) with specific constraints on data types and values.
Capability inventory: The skill utilizes numerical simulation, optimization, and local diagnostic capabilities via mfapy, scipy, and nlopt.
Sanitization: String identifiers are validated against a strict alphanumeric regex to prevent them from carrying executable instructions into the simulation engine.

13c-metabolic-flux