Genome Validate

Multi-agent validation pipeline for fact-checking claims, verifying evidence tiers, and ensuring prescriber document accuracy. Uses Codex CLI, NotebookLM, PubMed subagents, and Tavily search. Supports: audit (full vault), gene SYMBOL (single gene fact-check), protocol NAME (protocol/report fact-check). Triggers on: /genome-validate, "validate this note", "fact-check", "cross-check claims", "verify evidence", "audit this report", "fact-check gene X", "check protocol X".

43 stars 4 forksAI, LLMs & Agents

Install

skills CLI (Claude Code, Cursor, Codex and more)
npx skills add glebis/genome-toolkit --skill genome-validate
Manual install
git clone --depth 1 https://github.com/glebis/genome-toolkit.git /tmp/genome-toolkit
mkdir -p ~/.claude/skills
cp -r /tmp/genome-toolkit/skills/genome-validate ~/.claude/skills/genome-validate

Personal skills live in ~/.claude/skills. Use .claude/skills in a project folder to share a skill with your team.

From the SKILL.md

# Genome Validate

Multi-agent validation for genome vault claims and reports.

## Vault Configuration
- Agent config: `config/agents.yaml`
- Validation logic: `scripts/lib/multi_agent.py`
- Evidence tiers: `config/evidence_tiers.yaml`

## Validation Modes

### Mode 1: Single Note Validation
Validate a specific gene note or research finding.

1. Read the note content
2. Dispatch to configured agents in parallel:
   - **Codex CLI**: Review evidence tiers, effect sizes, drug interactions
   - **Claude subagent (PubMed)**: Search for recent publications on this gene
   - **Tavily**: Check for retractions of cited papers
3. Aggregate results using consensus logic
4. Report: pass/warn/block with specific flags

**Gate**: `gene_note` — requires 1 agent pass, blocks on effect_size_mismatch, wrong_evidence_tier

### Mode 2: Report Validation (Gated)
Validate prescriber-facing documents (Wallet Card, PGx Card, Prescriber Summary).

1. Read the report content
2. Dispatch to agents:
   - **Codex CLI**: Check drug contraindications against CPIC guidelines
   - **NotebookLM**: Verify claims against uploaded source papers
   - **Claude subagent**: Cross-reference drug interactions
3. Apply strict consensus gate
4. **BLOCK publication** if safety-critical errors found

**Gate**: `prescriber_report` — requires 2 agents agree, zero tolerance on drug safety

### Mode 3: Full Vault Audit
Comprehensive multi-agent audit of the entire vault.
Read the full skill on GitHub

Details

Skill name
genome-validate
Path
skills/genome-validate/SKILL.md
License
MIT
Repo updated
3 June 2026

FAQ

What does the Genome Validate skill do?

Multi-agent validation pipeline for fact-checking claims, verifying evidence tiers, and ensuring prescriber document accuracy. Uses Codex CLI, NotebookLM, PubMed subagents, and Tavily search. Supports: audit (full vault), gene SYMBOL (single gene fact-check), protocol NAME (protocol/report fact-check). Triggers on: /genome-validate, "validate this note", "fact-check", "cross-check claims", "verify evidence", "audit this report", "fact-check gene X", "check protocol X".

How do I install the Genome Validate skill?

skills CLI (Claude Code, Cursor, Codex and more): npx skills add glebis/genome-toolkit --skill genome-validate. Skills go in ~/.claude/skills (personal) or .claude/skills (project), and Claude loads them automatically when a task matches the description.

Is Genome Validate popular?

It lives in glebis/genome-toolkit, which has 43 GitHub stars and 4 forks.

Related skills

Genome Analytics

43

glebis/genome-toolkit

Run analysis scripts: PRS, enrichment, vault audit, PubMed monitoring, consistency checks. Triggers on: /genome-analytics, "run PRS", "audit vault", "check PubMed", "run enrichment", "check consistency", "analyze pathway".

Data & Analytics

Genome Create

43

glebis/genome-toolkit

Create gene, system, or phenotype notes from SQLite genotype data and templates. Triggers on: /new-gene SYMBOL, /new-system NAME, "add gene X", "create system note", "create phenotype note for X".

Skill & Plugin Authoring

Genome Import

43

glebis/genome-toolkit

Universal genome data import supporting 23andMe v4/v5, AncestryDNA, MyHeritage, Nebula, and generic VCF. Also handles imputation prep and post-imputation import. Triggers on: /genome-import, "import my DNA data", "add raw file", "prepare for imputation", "import imputed data".

Other

Inkbox Contact Rules

44

inkbox-ai/hermes-agent-plugin

Use when the user wants to block, allow, delete, or list Inkbox contact-rule filters for the agent's mailbox or phone number. Hermes does not expose contact-rule edit tools; guide the user to Inkbox Console or explain the limitation.

AI, LLMs & Agents10

Sanitize

19

agentward-ai/agentward

Detect and redact PII from text files. Supports 15 categories including credit cards, SSNs, emails, API keys, addresses, and more — with zero dependencies.

AI, LLMs & AgentsIn 1 list

Inkbox Outbound Calling

44

inkbox-ai/hermes-agent-plugin

Use when the user asks the agent to place an outbound Inkbox phone call, call a phone number/contact, or call someone with a specific purpose or opening message.

AI, LLMs & Agents10

Browse the most popular Claude skills, all ai, llms & agents skills, or write your own with the skills generator.