How it works.
Name who needs to decide
State who must decide and the time horizon in which the decision matters.
Map constraints and incentives
Use the evidence to identify resources, dependencies, feedback and competing goals.
Separate evidence from inference
Mark which relationships are observed and which are analytical judgments.
Compare choices
Describe which choices open, narrow, remain unchanged, or close, including the conditions that would reverse that result.
What it produces—and when it fails.
Output
An explanation of the available choices, what each depends on and what would change the assessment. It does not prescribe a choice.
Failure condition
It fails if it invents a decision holder, forces every situation into a phase model, presents an unsupported probability, or recommends action without showing the evidence chain.
Worked example
When infrastructure borrowing crosses a historical threshold, Manticus examines financing options and what assets might be worth to a future owner. It states which evidence would change that assessment.
A guide to reasoning, not a running model.
Manticus draws on systems theory, active inference, and the boundary concept often described through Markov blankets. FP1 uses these as reasoning prompts; the current workflow does not solve an active-inference model or calculate a Markov blanket.
A named reviewer checks AI-assisted work against its sources and these failure conditions before it enters a decision brief. That reviewer is responsible for the conclusion. See how the methods fit together.