CtrlK
BlogDocsLog inGet started
Tessl Logo

stat-result-validator

Validate statistical research outputs for formulation quality, method-to- problem alignment, theory presence, experimental evidence, fair comparison, artifact completeness, and final-claim consistency.

57

Quality

66%

Does it follow best practices?

Run evals on this skill

Adds up to 20 points to the overall score

View guide

SecuritybySnyk

Passed

No findings from the security scan

Fix and improve this skill with Tessl

tessl review fix ./external/agents/stat_research_agent/skills/stat-result-validator/SKILL.md
SKILL.md
Quality
Evals
Security

Stat Result Validator

Overview

Use this skill after formulation, method proposal, theory, experimental evaluation, comparison, and result synthesis. It checks whether the final result is supported by a coherent statistical research chain.

Artifact Checks

Required for all topics:

progress/<TOPIC_ID>/step0_problem_formulation.md
progress/<TOPIC_ID>/step1_method_proposal.md
progress/<TOPIC_ID>/step2_theory_analysis.md
progress/<TOPIC_ID>/step3_experimental_evaluation.md
progress/<TOPIC_ID>/step4_comparison.md
progress/<TOPIC_ID>/step5_result_synthesis.md
progress/<TOPIC_ID>/step6_quality_audit.md
experiments/<TOPIC_ID>/config.yaml
experiments/<TOPIC_ID>/results/metrics.json
experiments/<TOPIC_ID>/results/run_manifest.json
experiments/<TOPIC_ID>/results/comparison_summary.md
experiments/<TOPIC_ID>/results/claim_verdicts.json
experiments/<TOPIC_ID>/report/paper.md
experiments/<TOPIC_ID>/README.md

Analysis-specific source files and raw outputs are determined by the experiment plan and should live under experiments/<TOPIC_ID>/src/ and experiments/<TOPIC_ID>/results/.

Formulation Checks

The formulation must define:

  • Observed data and sampling regime
  • Data model or data source
  • Target parameter, decision, prediction, or risk
  • Assumptions
  • Claims or hypotheses
  • Evaluation criteria
  • Theory targets

Blocking failures:

  • No target or estimand.
  • Claims cannot be measured.
  • Assumptions are absent or incompatible with the proposed method.
  • Evaluation criteria do not answer the research question.

Method Checks

Verify that:

  • The proposed method addresses the formulated target.
  • Baselines are meaningful.
  • Ablations isolate important design choices.
  • Diagnostics are specified for likely failure modes.
  • Method outputs match the metrics and theory targets.

Theory Checks

Theory may be rigorous or partial, but it must be explicit.

Check for:

  • Definitions and assumptions.
  • Proposition, theorem, derivation, counterexample, or clearly labeled heuristic analysis.
  • Predicted empirical patterns.
  • Limitations and regimes not covered.

Blocking failures:

  • No theory section at all.
  • Theory analyzes a different target than the formulation.
  • Final claims are stronger than the theory supports.

Experimental Evidence Checks

Check that:

  • Experiments test the formulated claims.
  • Conditions are aligned with assumptions and stress tests.
  • Proposed method, baselines, and ablations are run on comparable conditions.
  • Failure counts are recorded.
  • Runtime reductions are recorded in run_manifest.json.
  • Metrics are grouped at the right level.

Comparison Checks

Verify that:

  • Comparisons include proposed method vs baseline.
  • Ablations are interpreted.
  • Experimental patterns are compared with theoretical predictions.
  • Disagreements between theory and experiments are discussed.
  • Claim verdicts cite both theoretical and empirical support when available.

Final Claim Checks

Every final claim must be traceable to:

formulation -> method -> theory -> experiment -> comparison

Use:

  • PASS: formulation, theory, experiments, and comparisons support the claims.
  • WARN: usable but has limitations that must be disclosed.
  • FAIL: missing formulation, theory, evidence, or fair comparison prevents a valid conclusion.
Repository
aiming-lab/AutoResearchClaw
Last updated
First committed

Is this your skill?

If you maintain this skill, you can claim it as your own. Once claimed, you can manage eval scenarios, bundle related skills, attach documentation or rules, and ensure cross-agent compatibility.