validation-agent

validation-agent is an agent for Claude Code from equinor/neqsim. It costs 38 tokens per session (2,899 once invoked), scanned A, original, Apache-2.0.

A statistical review agent for computational experiments that checks whether claimed improvements are supported by repeated results. It tests comparisons or checks coverage, consistency, cross-validation, and scaling for descriptive studies.

In plain words
What is it for?
Use it to analyze benchmark runs, test statistical significance, identify differing conditions, and produce validated claims and reports.
Why use it?
It prevents manuscripts from making unsupported claims about performance or improvement. Approved claims are separated from results that need qualification.

Agent for Claude Code

Written for Claude Code: a Claude Code subagent (agents/*.md).

Good fit Use it to analyze benchmark runs, test statistical significance, identify differing conditions, and produce validated claims and reports.

Compare 6 agents from other repositories ↓
Install with agentmods
npx agentmods add agents/equinor/neqsim/validation.paperlab
Install

Getting it into your agent

One page per mod, every tool's command on it. A separate URL per tool would split the same page into five that compete with each other.

Clone the repo
git clone --depth 1 https://github.com/equinor/neqsim

Made for: Claude Code.

Wrote this? Show the measurements

A badge with what this costs and how it scanned, read live from this page, so it follows the numbers instead of freezing them. Markdown for a README, HTML for a documentation site or a project page.

agentmods badge for validation-agent

README.md
[![agentmods](https://agentmods.dev/badge/agents/equinor/neqsim/validation.paperlab.svg)](https://agentmods.dev/agents/equinor/neqsim/validation.paperlab)
Your own site
<a href="https://agentmods.dev/agents/equinor/neqsim/validation.paperlab"><img src="https://agentmods.dev/badge/agents/equinor/neqsim/validation.paperlab.svg" alt="Measured on agentmods" height="20"></a>
Per session 38 Only the description is in the session, so the agent can decide to use it. The body loads when it is invoked.
When invoked 2,899 The whole file, excluding the scripts and references it only reads on demand.
Security scan A 0 findings. A grade says what 26 rules found in the file — not that it is safe.
Origin original No closer match found in the catalogue.
Token cost

What it costs to keep this loaded

Counted locally with the o200k_base tokenizer, which is exact for GPT models; Claude uses its own tokenizer and its counts differ. Treat this as one consistent yardstick across the catalogue rather than a bill. Prices are per million input tokens.

ModelPer sessionOnce invoked
Fable 5.1 $0.00038 $0.02899
Opus 5 $0.00019 $0.01450
Sonnet 5 $0.00008 $0.00580
Haiku 4.5 $0.00004 $0.00290

Measured 3d ago against content hash b524fce740e8, method: parsed. Prices are Anthropic first-party input rates as of 2026-09-07, from the pricing page.

Security

Grade A, and why

validation-agent scanned grade A with 0 findings against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured 3d ago.

A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.

Nothing flagged

None of the 26 patterns this scan looks for appear in this file: no shell pipes, no recursive deletes, no credential paths, no hidden text, no instruction-override or anti-refusal phrasing, no agent-config snooping. That is not a guarantee, it is the absence of the things that are checkable.

neqsim-paperlab/agents/validation.paperlab.md · 318 lines

How it starts

The opening of the file, as written. The whole thing — 318 lines — stays where its author put it; the contents beside it link to each section on GitHub.

Validation Agent

You are a statistical validation specialist for computational experiments. Your job is to determine which claims are supported by evidence and which are not.

Your Role

Given benchmark results and the paper type, you:

  1. Aggregate results across repeated runs
  2. Test statistical significance of improvements (Type 1) OR characterize distributions (Type 2)
  3. Identify regimes where behavior differs
  4. Produce approved_claims.json (Type 1) OR validate results.json completeness (Type 2)
  5. Write validation_report.md — full analysis

Validation by Paper Type

Type 1 (Comparative): Full Statistical Pipeline

Use paired tests (Wilcoxon, chi-square) with p < 0.05. Require effect sizes and confidence intervals. Gate claims through approved_claims.json.

Type 2 (Characterization): Completeness and Consistency

No A-vs-B comparison needed. Instead verify:

  • Coverage: All planned conditions were tested (no missing cases)
  • Consistency: Repeated runs give stable results (timing CV < 30%)
  • Cross-validation: EOS or solver cross-check agrees on phase ID
  • Scaling: Reported scaling relationship fits the data (R² > 0.8)
  • No silent failures: All exceptions caught and cataloged Produce a validation_report.md documenting these checks.

Type 3 (Method): Mathematical + Computational

  • Verify mathematical claims against analytical solutions
  • Check convergence order matches theoretical prediction
  • Validate against reference solutions from JANAF/NASA/NIST
  • Report element/mass balance closure for reactor problems

Type 4 (Application): External Reference Validation

  • Compare against published experimental or simulation data
  • Report AAD%, max deviation, and bias
  • Identify conditions where deviations exceed acceptable thresholds

Validation Framework

Claim Types

Type Example Validation Method Paper Types
Convergence improvement "Converges in 15% fewer cases" Chi-square test on convergence rates Type 1
Speed improvement "30% faster on average" Paired t-test or Wilcoxon signed-rank Type 1
Robustness improvement "Handles near-critical better" Subset analysis by fluid family Type 1
No regression "No slower on easy cases" One-sided test for non-inferiority Type 1
Coverage claim "100% convergence across 1664 cases" Count-based, no stat test needed Type 2
Scaling claim "Time scales as 0.015×Nc ms" Linear regression, report R² Type 2
Distribution claim "Median CPU time 0.088 ms" Descriptive statistics from data Type 2
Accuracy claim "AAD < 2% vs NIST data" Direct comparison against reference Type 3, 4
Mathematical claim "Quadratic convergence near solution" Convergence rate analysis Type 3

Read the full file on GitHub · 318 lines

Changes

What this file has done since we first saw it

Hashed on every crawl. A supply-chain change to an agent config is a question of when, not whether, so the history is kept rather than the latest state alone.

  1. 3d ago First seen · 318 lines · 38 tokens per session scan A b524fce740e8

Subscribe to this mod's changes

validation-agent is an agent published in the GitHub repository equinor/neqsim (150 stars, last pushed yesterday), licensed Apache-2.0. It adds 38 tokens to every session and 2,899 once invoked, about $0.0002 per session on Opus 5. A static security scan graded it A with 0 findings. No closer match exists in the catalogue, so it is treated as the original; first seen 2026-09-03.