| name | choose |
| description | Compare options and recommend a direction. Use for decision requests like "choose", "which option", "tradeoffs", "recommend", "should we", and option selection with criteria, risks, and reversibility. |
| license | MIT |
| tags | ["decision","recommendation","tradeoffs"] |
| metadata | {"author":"Oleg Shulyakov","version":"1.1.0","source":"github.com/olegshulyakov/agent.md","catalog":"utility","category":"project-management"} |
choose
Choose a direction by comparing viable options against explicit criteria.
Workflow
- Identify the decision, options, and constraints.
- Define criteria, prioritizing user-provided criteria over inferred ones.
- Remove non-viable options with brief reasons.
- Compare viable options against the criteria.
- Recommend a direction, including assumptions, risks, tradeoffs, and reversibility.
- Name what evidence would change the decision.
Output
- Lead with the recommendation: state the choice when the evidence supports one.
- Show the basis: include criteria and concise option comparison.
- Name tradeoffs: explain what the recommendation gives up.
- State reversibility: note whether the choice is easy to change later.
- Handle ties honestly: recommend a tie-breaker or next evidence step when options remain balanced.
Boundaries
Scenario: Criteria are needed for comparison
Given options are being compared
Then compare them against goals, constraints, risk, cost, speed, reversibility, maintenance, user impact, or user-provided criteria
Scenario: Evidence supports a recommendation
Given evidence is sufficient to prefer one option
Then choose one option
And state when the evidence is not sufficient
Scenario: Output drifts into classification
Given grouping options is useful
Then use grouping only as support for a decision
And keep the primary output focused on choosing
Error Paths
Scenario: No criteria are provided
Given the user has not provided decision criteria
Then infer practical criteria
And label them as assumptions
Scenario: No options are provided
Given the user has not provided options
Then define plausible options before comparing them
Scenario: Evidence is insufficient
Given the available evidence cannot support a firm recommendation
Then provide a conditional recommendation
And name the smallest information needed to firm it up
Verification
Scenario: Output passes quality check
Given a recommendation has been produced
Then the selected option wins on the criteria that matter most
And subjective preferences and uncertain assumptions are surfaced
And meaningful risks, mitigations, and reversibility are included