recipe-validate
Skill7 starsChanged 3 months agoMITView on GitHub
Validates a hypothesis with a risk-appropriate method and records decision-relevant evidence. Use when testing Value, Usability, Feasibility, or Viability assumptions.
In this repository15 other files it gives its agents.
- blueprint-standardsSkill
- business-contextSkill
- design-perspectiveSkill
- hypothesis-disciplineSkill
- prd-standardsSkill
- product-principlesSkill
- prototype-guideSkill
- recipe-blueprintSkill
- recipe-defineSkill
- recipe-discoverSkill
- recipe-personaSkill
- recipe-prototype-promptSkill
- recipe-refine-visualsSkill
- recipe-reflectSkill
- recipe-visionSkill
The file
We split this skill into the rules it gives an agent
11 rules. About 0 are common, 8 are distinctive, and 3 are very specific to this project.
≈0
Common · estimate
Many files like this have a rule of this kind.
≈8
Distinctive · estimate
This author's own choices. Worth a look.
3
Very specific
Tied to their project's names, files or tools. Probably skip.
- Common
- Distinctive
- Very specific
- Very specific: line 18
WHEN the selected method is prototype validation: [LOAD IF NOT ACTIVE] prototype-guide
- Very specific: line 19
WHEN the selected method uses business-model, market, or viability analysis: [LOAD IF NOT ACTIVE] business-context
- Distinctive: line 29
Read the target hypothesis file(s).
- Distinctive: line 49
When any of these conditions applies, end the current turn with the validation design as the workflow output.
- Very specific: line 75
Use web search to gather market data.
- Distinctive: line 81
Update the hypothesis file:
- Distinctive: line 83
Update confidence scores with evidence
- Distinctive: line 85
Document a learning when the result can change a later decision
- Distinctive: line 96
When a hypothesis with a deadline reaches it without conclusion:
- Distinctive: line 99
Extend: Add more time with justification
- Distinctive: line 101
Abandon: Stop validation, record learnings from partial evidence
≈ “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.
Files
Context: Validate hypotheses using appropriate methods based on risk type. Invoke hypothesis-verifier for bias-free validation design. Record results in hypothesis files.
Required Skills [LOAD BEFORE EXECUTION]
- [LOAD IF NOT ACTIVE]
hypothesis-discipline— lifecycle, evidence, confidence, and stopping conditions - [LOAD IF NOT ACTIVE]
product-principles— 4 Risks and validation sufficiency
Conditional Skills [LOAD WHEN TRIGGERED]
- WHEN the selected method is prototype validation: [LOAD IF NOT ACTIVE]
prototype-guide - WHEN the selected method uses business-model, market, or viability analysis: [LOAD IF NOT ACTIVE]
business-context
Delegate validation design to hypothesis-verifier for bias-free assessment.
Execution Decision Flow
1. Hypothesis Assessment
Input: Use the path or text supplied with the explicit skill invocation. If no input was supplied and the target cannot be inferred unambiguously, ask for it.
Read the target hypothesis file(s). Understand:
- What is being tested (the hypothesis statement)
- Which risk dimension is primary (Value / Usability / Feasibility / Viability)
- Current confidence scores
- Validation stopping condition, including a time budget or deadline when present
- Parent Opportunity context
2. Validation Design
Invoke hypothesis-verifier to design the validation:
- hypothesis-verifier operates in a separate context to prevent confirmation bias
- It designs the test without knowing the orchestrator's expectations
- It defines success/failure criteria independently
Present the validation design to the user when executing it would commit external resources, change the product outcome, or exceed the confirmed time or cost boundary:
- Proposed validation method
- Success and failure criteria
- Required resources and time estimate
- Risk of the validation approach itself
When any of these conditions applies, end the current turn with the validation design as the workflow output. Execute that validation only after the user confirms it in a later turn.
Otherwise, execute local and reversible validation within the confirmed boundary without an additional stop.
3. Validation Execution
Execute validation based on the risk type and method:
| Risk Type | Validation Methods | Tools |
| Value | Market research, user interviews, competitive analysis, landing page test | Web search, survey analysis |
| Usability | Prototype testing, usability study, interaction analysis | prototype-generator (per prototype-guide skill) |
| Feasibility | Code spike, architecture review, dependency analysis | codebase-analyzer, code execution |
| Viability | Business model analysis, ROI calculation, regulatory review | Web search, BMC/VPC analysis |
Prototype Generation (for Usability validation)
Invoke prototype-generator in a separate context:
- Pass
hypothesis_pathas the canonical target hypothesis path. - Pass one exact
output_pathper invocation. Usedocs/discovery/prototypes/hypo-{id}-prototype.htmlby default; when the confirmed validation design requires separate artifacts, usedocs/discovery/prototypes/hypo-{id}-{variant}-prototype.htmlfor each artifact. - The generator reads the decision-relevant product and design sources directly and writes one self-contained HTML artifact at that path. The validation design determines whether one artifact or separate variant artifacts supply sufficient evidence.
- Consume its completion report as prototype-readiness evidence. When it returns
blocked, resolve only the named incomplete condition before invoking it again. - Collect the observations defined by the confirmed validation design before recording a hypothesis result. When observation has not run, report the prototype as ready and validation as pending; use
inconclusiveonly when executed validation produced insufficient evidence. - Continue validation and result recording in the parent context; prototype-generator does not update the hypothesis record.
Market Research (for Value/Viability validation)
Use web search to gather market data. See product-principles skill references/mvp-definition.md for scope assessment.
4. Result Recording
After validation execution:
- Update the hypothesis file:
- Change status (validated / invalidated / inconclusive / timeout)
- Update confidence scores with evidence
- Record evidence and artifacts
- Document a learning when the result can change a later decision
- Link to validation artifacts (prototypes, data, interview notes)
Present the results to the user:
- Validation outcome (validated / invalidated / inconclusive)
- Updated confidence scores with evidence
- Key learnings
- Next decision, when the result creates one
5. Timeout Handling
When a hypothesis with a deadline reaches it without conclusion:
- Set status to
timeout - Present options to user:
- Extend: Add more time with justification
- Pivot: Modify the hypothesis based on partial evidence
- Abandon: Stop validation, record learnings from partial evidence
Sub-agent Usage
| Agent | When | Why (context separation benefit) |
| hypothesis-verifier | Always (validation design) | Eliminates confirmation bias in test design |
| prototype-generator | Prototype validation | Keeps generated frontend code and visual iteration out of validation design and result-recording context |
| codebase-analyzer | Feasibility validation | Objective code analysis without hypothesis bias |
Scope Boundaries
Included: Hypothesis validation design, validation execution, result recording Not included: Hypothesis generation (→ recipe-discover), PRD creation (→ recipe-define), reflection/distillation (→ recipe-reflect)
Completion
The workflow is complete when the hypothesis record contains the validation result and evidence, including a bounded inconclusive or timeout result, and the user has received the next decision when one exists.
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
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.

