agentleFS
TrecekTrecek/useful-claude-skills/.claude/skills/exp-lens-exploratory-confirmatory/SKILL.md

exp-lens-exploratory-confirmatory

Skill1 starsChanged 8 months agoMITView on GitHub

Assess whether analytic decisions were pre-specified or post-hoc and whether exploratory/confirmatory norms are aligned. Boundary lens answering "Is this discovery or test, and are norms aligned?

The file

We split this skill into the rules it gives an agent

11 rules. About 0 are common, 9 are distinctive, and 2 are very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈9

    Distinctive · estimate

    This author's own choices. Worth a look.

  • 2

    Very specific

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

  • Common
  • Distinctive
  • Very specific
  1. Distinctive:

    Do not litter the codebase with useless comments, TODO markers, or explanatory annotations — the skill output and diagram speak for …

    line 30
  2. Very specific:

    BEFORE creating any diagram, LOAD the /mermaid skill using the Skill tool - this is MANDATORY

    line 37
  3. Distinctive:

    Look for: preregister, protocol, plan, a_priori, predetermined, registered, osf

    line 49
  4. Distinctive:

    Look for: choice, option, alternative, try, different, also, additionally, robustness

    line 53
  5. Distinctive:

    Look for: main_result, primary, secondary, supplementary, appendix, not_shown, data_not_shown

    line 57
  6. Distinctive:

    Look for: we found, unexpectedly, interestingly, surprising, post_hoc, exploratory, we noticed

    line 61
  7. Distinctive:

    Look for: holdout, validation, train_dev_test, discovery_set, confirmation_set, split

    line 65
  8. Distinctive:

    If not, how many alternative analyses could have been run?

    line 81
  9. Distinctive:

    Use the mermaid skill conventions with:

    line 90
  10. Very specific:

    Write the output to: temp/exp-lens-exploratory-confirmatory/exp_diag_exploratory_confirmatory_{YYYY-MM-DD_HHMMSS}.md

    line 107
  11. Distinctive:

    Before creating the diagram, verify:

    line 209

≈ “Common” is an estimate: at least 1 in 5 of the 372,429 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

Exploratory-Confirmatory Experimental Design Lens

Philosophical Mode: Boundary Primary Question: "Is this discovery or test, and are norms aligned?" Focus: Pre-specification, Analytic Flexibility, HARKing Detection, Garden of Forking Paths, Transparent Reporting

When to Use

  • Study mixes exploration and confirmation without clear boundaries
  • Post-hoc hypotheses presented as pre-specified
  • Many analyses run but only significant ones reported
  • User invokes /exp-lens-exploratory-confirmatory or /make-experiment-diag exploratory

Critical Constraints

NEVER:

  • Modify any source code files
  • Do not litter the codebase with useless comments, TODO markers, or explanatory annotations — the skill output and diagram speak for themselves

ALWAYS:

  • Map the full analytic timeline — what was decided before vs. after seeing data
  • Count forking paths honestly — every analysis choice is a potential fork
  • Distinguish genuine exploration (hypothesis-generating) from HARKing (hypothesis-after-results)
  • Flag absent preregistration as a finding without assuming bad faith
  • BEFORE creating any diagram, LOAD the /mermaid skill using the Skill tool - this is MANDATORY

Analysis Workflow

Step 1: Launch Parallel Exploration Subagents

Spawn Explore subagents to investigate:

Pre-specified Plans

  • Find evidence of pre-registration, protocols, or analysis plans written before seeing data
  • Look for: preregister, protocol, plan, a_priori, predetermined, registered, osf

Analytic Flexibility

  • Find how many analysis choices could have been made differently
  • Look for: choice, option, alternative, try, different, also, additionally, robustness

Selective Reporting Signals

  • Find whether all planned analyses are reported
  • Look for: main_result, primary, secondary, supplementary, appendix, not_shown, data_not_shown

Post-Hoc Rationalization

  • Find hypotheses that appear to have been generated from the data
  • Look for: we found, unexpectedly, interestingly, surprising, post_hoc, exploratory, we noticed

Exploration-Confirmation Separation

  • Find whether there is an explicit boundary between exploration and test
  • Look for: holdout, validation, train_dev_test, discovery_set, confirmation_set, split

Step 2: Map Analytic Timeline

Map the analytic timeline:

  • What was decided before seeing data?
  • What was decided after?
  • Where is the boundary between exploration and confirmation?

Count the "forking paths" — decision points where a different choice would have led to a different result.

Step 3: Analyze Boundary Integrity

CRITICAL — Analyze Boundary Integrity: For every reported result:

  • Was the analysis plan fixed before seeing the outcome?
  • If not, how many alternative analyses could have been run?
  • Does the reporting distinguish exploratory findings from confirmatory tests?

Assess whether the reported results have survivorship bias.

Step 4: Create the Diagram (Optional)

This is primarily an argumentative lens. Create an optional timeline diagram only when the timeline is complex enough to benefit from visualization.

Use the mermaid skill conventions with:

Direction: LR (time flows left to right from planning through analysis to reporting)

Minimal diagram structure:

  • Pre-data decisions → Data collection → Post-data decisions → Reporting

Node Styling:

  • stateNode class: Pre-specified decisions
  • handler class: Data collection
  • phase class: Post-data decisions
  • gap class: Undisclosed flexibility
  • detector class: Exploration/confirmation boundary
  • output class: Reported results

Step 5: Write Output

Write the output to: temp/exp-lens-exploratory-confirmatory/exp_diag_exploratory_confirmatory_{YYYY-MM-DD_HHMMSS}.md


Output Template

# Exploratory-Confirmatory Analysis: {Experiment Name}

**Lens:** Exploratory-Confirmatory (Boundary)
**Question:** Is this discovery or test, and are norms aligned?
**Date:** {YYYY-MM-DD}
**Scope:** {What was analyzed}

## Analytic Timeline

| Decision Point | Pre/Post Data | Flexibility | Impact on Results |
|---------------|---------------|-------------|-------------------|
| {decision} | {Pre/Post} | {Low/Medium/High} | {description} |

## Forking Paths Inventory

| Analysis Choice | Alternatives Not Taken | Estimated Impact |
|----------------|------------------------|-----------------|
| {choice made} | {alternative 1, 2...} | {Low/Medium/High} |

## HARKing Risk Assessment

| Finding | Framing | HARKing Indicators | Assessment |
|---------|---------|-------------------|------------|
| {finding} | {how presented} | {indicators found} | {Low/Medium/High risk} |

## Timeline Diagram (Optional)

```mermaid
%%{init: {'flowchart': {'nodeSpacing': 50, 'rankSpacing': 60, 'curve': 'basis'}}}%%
graph LR
    %% CLASS DEFINITIONS %%
    classDef cli fill:#1a237e,stroke:#7986cb,stroke-width:2px,color:#fff;
    classDef stateNode fill:#004d40,stroke:#4db6ac,stroke-width:2px,color:#fff;
    classDef handler fill:#e65100,stroke:#ffb74d,stroke-width:2px,color:#fff;
    classDef phase fill:#6a1b9a,stroke:#ba68c8,stroke-width:2px,color:#fff;
    classDef newComponent fill:#2e7d32,stroke:#81c784,stroke-width:2px,color:#fff;
    classDef output fill:#00695c,stroke:#4db6ac,stroke-width:2px,color:#fff;
    classDef detector fill:#b71c1c,stroke:#ef5350,stroke-width:2px,color:#fff;
    classDef gap fill:#ff6f00,stroke:#ffa726,stroke-width:2px,color:#000;
    classDef integration fill:#c62828,stroke:#ef9a9a,stroke-width:2px,color:#fff;

    PRESPEC["Pre-specified Decisions<br/>━━━━━━━━━━<br/>Hypothesis, primary<br/>outcome, analysis plan"]

    COLLECT["Data Collection<br/>━━━━━━━━━━<br/>Observation period<br/>and data gathering"]

    BOUND["E/C Boundary<br/>━━━━━━━━━━<br/>Explicit split between<br/>exploration and test"]

    POSTDATA["Post-data Decisions<br/>━━━━━━━━━━<br/>Analysis choices made<br/>after seeing results"]

    HIDDEN["Undisclosed Flexibility<br/>━━━━━━━━━━<br/>Unreported forks<br/>and alternatives"]

    REPORTED["Reported Results<br/>━━━━━━━━━━<br/>Published findings<br/>and conclusions"]

    PRESPEC --> COLLECT
    COLLECT --> BOUND
    BOUND --> POSTDATA
    POSTDATA --> HIDDEN
    POSTDATA --> REPORTED
    HIDDEN -.->|"survivorship bias"| REPORTED

    %% CLASS ASSIGNMENTS %%
    class PRESPEC stateNode;
    class COLLECT handler;
    class BOUND detector;
    class POSTDATA phase;
    class HIDDEN gap;
    class REPORTED output;

Preregistration Compliance Checklist

  • Hypothesis stated before data collection
  • Primary outcome pre-specified
  • Analysis plan documented before analysis
  • Deviations from plan disclosed
  • Exploratory analyses labelled as such

Transparent Reporting Scorecard

CriterionMet?Notes
All planned analyses reported{Yes/No/Partial}{notes}
Exploratory/confirmatory distinction explicit{Yes/No/Partial}{notes}
Forking paths disclosed{Yes/No/Partial}{notes}
Null results reported{Yes/No/Partial}{notes}

Key Findings

  • {Description of boundary integrity assessment and most significant HARKing or flexibility risks}

---

## Pre-Diagram Checklist

Before creating the diagram, verify:

- [ ] LOADED `/mermaid` skill using the Skill tool
- [ ] Using ONLY classDef styles from the mermaid skill (no invented colors)
- [ ] Diagram will include a color legend table

---

## Related Skills

- `/make-experiment-diag` - Parent skill for lens selection
- `/mermaid` - MUST BE LOADED before creating diagram
- `/exp-lens-severity-testing` - For statistical severity and error probability analysis
- `/exp-lens-sensitivity-robustness` - For robustness of conclusions across analytic choices

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.