Skip to main content

Documentation Index

Fetch the complete documentation index at: https://feasible-1447f9c5.mintlify.app/llms.txt

Use this file to discover all available pages before exploring further.

​
Asking for clarification (interrupts)

When Feasible isn’t sure about something — an ambiguous constraint, a missing piece of data, a target that could be a goal or a hard limit — it will pause and ask. You see the question in the chat. Reply, and the workflow resumes. Nothing is lost. This happens at a few specific points in the workflow:
  • While building the standard form, when the problem statement is incomplete or ambiguous.
  • Before self-evaluation, in interactive mode — Feasible may ask “Want me to check my own work?”, useful when you can verify the answer yourself faster.

​
Steering the agent

You can guide Feasible by replying in natural language:
  • “Use big-M = 10000 for the indicator constraints.”
  • “Add a symmetry-breaking constraint: queen in row 1 must be in column ≤ 4.”
  • “That constraint is wrong — the cap is per quarter, not per month.”
  • “Re-solve with the gap tolerance set to 0.01.”
You can also re-run a thread with a different LLM provider — useful when one provider’s formulation is buggy and you want a second opinion.
If you’re an OR practitioner, see steering the agent effectively for patterns that work well — naming the formulation pattern, pre-computing data tables, specifying variable types upfront.