The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill defines several Python functions and a solver class intended to process structured network data (nodes, supplies, costs, and capacities) which could be sourced from untrusted external inputs.\n
Ingestion points: The solve_min_cost_flow_nx, solve_transportation_problem, and solve_multi_commodity_flow functions in SKILL.md accept data structures that could be populated via agent-extracted user data.\n
Boundary markers: The instructions do not specify any delimiters or safety markers to differentiate between instructions and data when interpolating untrusted content into the optimization functions.\n
Capability inventory: The skill utilizes NetworkX for graph algorithms and PuLP for invoking linear programming solvers. It also uses matplotlib for generating visual output.\n
Sanitization: There is no evidence of validation, schema enforcement, or data sanitization for the inputs before they are passed to the mathematical solvers or visualization functions.

network-flow-optimization