agentleFS
shinprshinpr/nautilus/.agents/skills/hypothesis-discipline/SKILL.md

hypothesis-discipline

Skill7 starsChanged 3 months agoMITView on GitHub

Manages hypothesis lifecycle, enforces validation criteria, time budgets, and confidence scoring rules. Use when creating hypotheses, updating confidence scores, setting validation criteria, handling timeouts, or recording validation results.

The file

We split this skill into the rules it gives an agent

4 rules. About 0 are common, 3 are distinctive, and 1 is very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈3

    Distinctive · estimate

    This author's own choices. Worth a look.

  • 1

    Very specific

    Tied to their project's names, files or tools. Probably skip.

  • Common
  • Distinctive
  • Very specific
  1. Distinctive:

    Add a deadline when a calendar cutoff changes the decision

    line 52
  2. Very specific:

    When deadline passes without conclusion → status becomes timeout

    line 53
  3. Distinctive:

    When a hypothesis reaches a conclusion, record the result, changed confidence dimensions, and evidence in the hypothesis file.

    line 66
  4. Distinctive:

    Inconclusive is honest: When evidence is insufficient, say so instead of forcing a verdict

    line 75

≈ “Common” is an estimate: at least 1 in 5 of the 364,980 skill files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.

Keep what fits and drop the rest. Your version, ready for your agents.
Files
SKILL.md

Hypothesis Discipline

Core Principle

A hypothesis is not a guess. A hypothesis selected for validation needs a clear decision, evidence criteria, and a stopping condition.

Hypothesis Characteristics

  • Hypotheses exist at every level of the OST hierarchy (see product-principles skill)
  • Each hypothesis has a target level attribute indicating which OST level it addresses
  • Hypotheses follow an ADR-style lifecycle — a single file tracks the full journey from draft to conclusion
  • Preserve rejected and invalidated hypotheses as learning assets

Hypothesis Lifecycle

draft → testing → validated → adopted
                            → rejected (validated but not adopted)
              → invalidated (disproven by evidence)
              → inconclusive (evidence gathered but insufficient to confirm or deny)
              → timeout (deadline passed, decision needed: continue or stop)

Hypothesis File Schema

The authoritative schema is defined in references/hypothesis-template.md. Key fields:

  • id: HYPO-NNN
  • level: outcome / opportunity / solution / assumption
  • status: draft / testing / validated / invalidated / inconclusive / adopted / rejected / timeout
  • confidence: scores for the risk dimensions that can change the hypothesis decision, on the product-principles 0-10 scale
  • time-budget and deadline: include when time or calendar limits affect the validation decision

Validation Criteria Requirements

A hypothesis entering validation defines:

  1. We believe that — the hypothesis statement
  2. We'll know we're right when — measurable success criteria
  3. We'll know we're wrong when — measurable failure criteria
  4. Validation method — how we will test (prototype, data analysis, interview, code spike, market research)
  5. Stopping condition — the evidence, time limit, or cost limit that ends this validation

Time Budget and Cutoff

  • Give a validation a time budget when exploration or evidence collection can expand
  • Add a deadline when a calendar cutoff changes the decision
  • When deadline passes without conclusion → status becomes timeout
  • Timeout forces a decision: extend (with justification), pivot, or abandon
  • End validation at its evidence, time, or cost boundary

Confidence Update Rules

  • Confidence scores are updated only when new evidence is gathered
  • Record the evidence that justified each score change
  • Confidence can go down as well as up — negative evidence is valid evidence
  • Different risk dimensions can have different confidence levels

Result Recording

When a hypothesis reaches a conclusion, record the result, changed confidence dimensions, and evidence in the hypothesis file. Carry a learning or parent-Opportunity update into reflection only when the result can change a later decision.

Key Disciplines

  • Separate creation from evaluation: Generate candidates before applying evaluation criteria
  • Seek disconfirming evidence: Actively look for reasons the hypothesis might be wrong
  • One decision per validation: Combine hypotheses when the same evidence resolves the same decision; separate them when bundling would make the result uninterpretable
  • Record decision-relevant evidence: Preserve the evidence and reasoning needed to understand or reuse the conclusion
  • Rejected ≠ worthless: A rejected hypothesis teaches what doesn't work and why
  • Inconclusive is honest: When evidence is insufficient, say so instead of forcing a verdict

Download now (ZIP)

Does it work?

Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.

No reviews yet

People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.

  • It worked
  • Partly
  • It didn't
What did you use it for? e.g. Reviewing pull requests in a Rails app

Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.

Your trail

Files you open will collect here, so you can find them again.

Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.

Now do this for your own.

Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.