Documentation Index
Fetch the complete documentation index at: https://docs.feasible.cloud/llms.txt
Use this file to discover all available pages before exploring further.
Things that help
- State the structure explicitly. “This is a flow problem on a DAG with 5 nodes” or “This is a packing problem — bins of capacity C, items of size s_i” gives the agent a strong prior.
- Specify variable types upfront. “Use binary indicator variables for assignments” or “Use continuous variables in [0,1] and round” sets the formulation direction.
- Pre-compute hard data. If a parameter is the result of a non-trivial calculation (compounding, NPV, etc.), pre-compute it and supply the table — keep the model linear.
- Be explicit about objectives. “Lexicographic: minimize cost first, then maximize fairness” or “weighted sum with weights 0.7 / 0.3.”
- Name the formulation pattern. “Use a flow-conservation formulation” or “Use a Miller-Tucker-Zemlin formulation for subtour elimination” — the agent recognizes these.
Things that hurt
- Vague objectives. “Optimize the schedule” — optimize what?
- Burying constraints in narrative. “We’ve always done it this way” doesn’t tell the agent the constraint.
- Implicit feasibility assumptions. State edge cases.