Use when a plan, decision, review, or research thread needs explicit hypotheses, evidence quality, measurement, or experiment design.
Adds engineering and scientific rigor to strategic work. Use this skill when the problem is not just "what should we say?" but "what do we actually know, what are we assuming, and how would we learn fast enough to change course?"
Rate evidence before leaning on it:
When evidence is mixed, say so.
For meaningful uncertainty, capture:
## Experiment Log
**Hypothesis:** {what we believe}
**Intervention:** {what we will test or change}
**Expected Observation:** {what would support or refute it}
**Actual Observation:** {what happened}
**Conclusion:** {continue, pivot, stop, or defer}
The point is not ceremony. The point is to keep the work falsifiable and learnable.
Before acting, ask:
Dead ends, failed tests, and weak signals are still evidence. Write them down so they stop being rediscovered as "new" ideas.
Escalate to an explicit matrix when:
Use rows such as:
/maestro:plan uses this skill for high-uncertainty plans and
bets/maestro:research uses it to rate evidence quality and name
what remains uncertain/maestro:review uses it to audit claims, success criteria, and
hidden assumptions/maestro:run uses it when the deliverable is an experiment,
memo, or evidence-backed recommendation/maestro:decide pairs with this skill when the decision should
include what evidence would reopen ittemplates/experiment-log-template.md — compact experiment
section for plans, reviews, or research notesexamples/experiment-note-example.md — worked example of a
hypothesis turning into a decision