10 concise reference notes for planning. These are public reference material, not private user records or a transfer of language-model weights.

State and goals

Describe the current state and the target state separately. An action has prerequisites, effects and a cost under the chosen model. A plan is a sequence of such actions; generating it does not perform those actions.

Authored reference note; not externally verified product advice · ref-041

Critical path

In a dependency schedule with stated resource assumptions, the longest prerequisite chain determines the earliest possible completion. Shortening a task outside the limiting chain may not shorten the overall schedule. Resource constraints can change the result.

Authored reference note; not externally verified product advice · ref-042

Dependency cycles

If task A requires B and B requires A without an already satisfied starting condition, the dependency model is cyclic. Report the cycle and change the model or prerequisites instead of generating an impossible ordering.

Authored reference note; not externally verified product advice · ref-043

Optimization scope

An optimizer finds the best option only within the supplied feasible set and objective. Missing alternatives, wrong costs or omitted constraints can make a mathematically optimal answer practically poor. Keep model assumptions visible.

Authored reference note; not externally verified product advice · ref-044

Pareto dominance

An alternative is dominated when another is at least as good on every considered criterion and strictly better on at least one, using the stated directions. This comparison depends on which criteria were included.

Authored reference note; not externally verified product advice · ref-045

Sensitivity analysis

Change important assumptions across plausible values and inspect whether the preferred result changes. A decision that reverses under a tiny input change needs more caution than one that remains preferred across a broad range.

Authored reference note; not externally verified product advice · ref-046

Reversible experiments

When uncertainty is high and the downside can be contained, use a small reversible experiment to gather evidence. Define the stopping condition and what would count as success before scaling the action.

Authored reference note; not externally verified product advice · ref-047

Expected information gain

A diagnostic test can be ranked by how much it is expected to reduce uncertainty under specified hypotheses and likelihoods. Information gain is not the same as business value, safety or a complete test-sequence optimum.

Authored reference note; not externally verified product advice · ref-048

Cost of delay

Compare implementation effort with the consequences of postponement. A blocking dependency can have high value even if it is not visually prominent. State how impact and urgency were estimated rather than hiding assumptions behind a priority score.

Authored reference note; not externally verified product advice · ref-049

Acceptance criteria

Define completion in terms of observable behavior. Include the expected result, relevant constraints and a verification method. Finishing a list of activities is not sufficient if the intended user outcome remains unavailable.

Authored reference note; not externally verified product advice · ref-050

In chat, use knowledge: topic. Browse all reference topics.