cross-validate

cross-validate is a command for coding agents from mims-harvard/ToolUniverse. It costs 59 tokens per session (1,209 once invoked), scanned A, original, Apache-2.0.

A command for checking one scientific claim against at least three independent databases and reporting whether the sources agree. It breaks a claim into testable parts before checking it.

In plain words
What is it for?
Use it to verify facts before publishing, citing, or acting on them, especially claims about research, medicines, variants, or other scientific data.
Why use it?
It reduces the risk of relying on one database or on several tools that all use the same underlying source.

Command

Part of the tooluniverse plugin — 140 skills, 8 commands, 1 agent, 1 hook, 1 MCP server shipped together

About the project

ToolUniverse is a collection of tools, interfaces, and supporting components for building AI systems that perform scientific work. It is for developers creating AI scientist agents that use APIs, databases, machine-learning tools, and domain-specific utilities. The catalogue includes skills, commands, an MCP server, an agent, and a hook for working with the ecosystem.

mims-harvard/ToolUniverse · 1,672 stars · on GitHub

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.

agentmods
npx agentmods add commands/mims-harvard/tooluniverse/cross-validate
Clone the repo
git clone --depth 1 https://github.com/mims-harvard/ToolUniverse

Or install tooluniverse, the plugin that ships this one along with the rest of its 140 skills, 8 commands, 1 agent, 1 hook, 1 MCP server.

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 cross-validate

README.md
[![agentmods](https://agentmods.dev/badge/commands/mims-harvard/tooluniverse/cross-validate.svg)](https://agentmods.dev/commands/mims-harvard/tooluniverse/cross-validate)
Your own site
<a href="https://agentmods.dev/commands/mims-harvard/tooluniverse/cross-validate"><img src="https://agentmods.dev/badge/commands/mims-harvard/tooluniverse/cross-validate.svg" alt="Measured on agentmods" height="20"></a>
Per session 59 Only the description is in the session, so the agent can decide to use it. The body loads when it is invoked.
When invoked 1,209 The whole file, excluding the scripts and references it only reads on demand.
Security scan A 0 findings. Scan, not verified.
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 $0.00059 $0.01209
Opus 5 $0.00030 $0.00605
Sonnet 5 $0.00012 $0.00242
Haiku 4.5 $0.00006 $0.00121

Measured 5d ago against content hash a12ee3b9772a, method: parsed. Prices are Anthropic first-party input rates as of 2026-08-30, from the pricing page.

Security

Grade A, and why

cross-validate 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 5d 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.

plugin/commands/cross-validate.md · 112 lines

How it starts

The opening of the file, as written. The whole thing — 112 lines — stays where its author put it; the contents beside it link to each section on GitHub.

Cross-validate this claim against independent databases: $ARGUMENTS

A single source for a high-stakes fact is fragile. Force 3+ sources from DIFFERENT databases (not 3 tools that all wrap the same upstream API).

Process

1. Decompose the claim into testable assertions

A claim like "BRAF V600E is FDA-approved indication for vemurafenib in melanoma" has three independent assertions:

  • (a) BRAF V600E exists as an annotated cancer variant
  • (b) Vemurafenib is FDA-approved
  • (c) The approved indication includes BRAF-V600E-positive melanoma

Each assertion can be confirmed/refuted separately. Write them out before calling tools — this is the anchor for what you'll verify.

For a numeric claim ("X% of breast cancers carry TP53 mutations"), the assertion is the value itself; cross-validate by getting the same number from independent cohorts.

2. Pick 3 independent databases per assertion

"Independent" means different upstream maintainers, not different tools wrapping the same API. Examples:

Domain Independent sources
Drug approvals OpenFDA, ChEMBL, DrugBank — each maintains its own approval table
Variant pathogenicity ClinVar (NIH), CIViC (academic crowd-sourced), OncoKB (MSKCC), gnomAD frequency (Broad)
Gene-disease association OpenTargets (composite), DisGeNET, OMIM, GenCC
Mutation frequency in cancer TCGA (GDC), COSMIC, IntOGen — each samples different cohorts
Clinical trial existence ClinicalTrials.gov, EU CTR, ChiCTR
Protein function UniProt, Reactome, KEGG
Pharmacology DrugBank, ChEMBL, PharmGKB, FDA Label

Pick 3. If the claim's domain only has 2 truly-independent sources, note that in the report — don't fabricate a third.

3. Run each source independently

Use tu run <tool> per source. Don't shortcut to a compound tool that aggregates multiple sources — that defeats the cross-validation point. Track each source's verdict separately:

Source 1 (ClinVar): CONFIRMS — V600E listed as Pathogenic, condition: Melanoma
Source 2 (CIViC):   CONFIRMS — V600E linked to vemurafenib sensitivity in melanoma
Source 3 (OncoKB):  CONFIRMS — V600 variant Level 1 evidence for vemurafenib in melanoma

Read the full file on GitHub · 112 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. 5d ago First seen · 112 lines · 59 tokens per session scan A a12ee3b9772a

Subscribe to this mod's changes

cross-validate is a command published in the GitHub repository mims-harvard/ToolUniverse (1,672 stars, last pushed today), licensed Apache-2.0. It adds 59 tokens to every session and 1,209 once invoked, about $0.0003 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 2026-08-30.