brain-in-the-fish

Decomposes documents into evidence-based verdicts using ontology-scored cognitive AI agents.

83|15|Updated Mar 23, 2026
One-click install
npx skills add https://github.com/fabio-rovai/brain-in-the-fish --skill brain-in-the-fish
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: brain-in-the-fish
Source: https://github.com/fabio-rovai/brain-in-the-fish/tree/main
Command: npx skills add https://github.com/fabio-rovai/brain-in-the-fish --skill brain-in-the-fish

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

Universal document evaluation engine — evaluate any document against any criteria using cognitively-modelled AI agents with ontology-grounded scoring

Core Features & Use Cases

  • Deterministic evidence scoring: each claim is backed by quotes or citations from the document.
  • Agent-ontology evaluation: a panel of AI agents maintains state as OWL ontologies to reason about sections and criteria.
  • Pipeline orchestration: ingest documents, load criteria, align sections, spawn evaluators, score, debate, and generate a final report.
  • Use Case: evaluate policy proposals, clinical reports, or tender responses with transparent justification and auditability.

Quick Start

Provide a sample document and run the full evaluation workflow to generate a score, ontology, and verdict.

Frequently Asked Questions about brain-in-the-fish

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I automate document evaluation against custom criteria?

Automate document evaluation by decomposing content against criteria with cognitive AI agents and ontology-grounded scoring. The modular pipeline ingests documents, aligns sections, spawns evaluators, scores evidence, debates findings, and generates a final verdict report.

What is ontology-grounded evidence scoring for document evaluation?

Ontology-grounded evidence scoring uses AI agents that maintain state as OWL ontologies to reason about document sections and criteria. Each evaluated claim is backed by direct quotes or citations, ensuring transparent justification and auditability for clinical or policy documents.

How do I evaluate policy proposals or clinical reports with transparent AI justification?

Evaluate policy proposals or clinical reports by running the full pipeline workflow on a sample document. The system applies deterministic scoring or Claude-assisted subagents to generate a score, ontology state, and final verdict with transparent justification.

Can I use SPARQL and OWL ontologies to reason about document sections?

Yes, a panel of AI agents maintains state as OWL ontologies to reason about sections and evaluation criteria. This agent-ontology evaluation approach supports deterministic evidence scoring and structured debate across the evaluation pipeline.

Does this document evaluation pipeline support deterministic scoring and Claude-assisted subagents?

The pipeline supports both deterministic scoring and Claude-assisted subagents for document evaluation. It enforces a frontmatter-driven entry point and integrates optional scripts, references, and assets while orchestrating ingest, criteria loading, alignment, scoring, and reporting.

When should I use ontology-driven document evaluation instead of manual review?

Use ontology-driven document evaluation for essays, policies, contracts, and clinical reports requiring evidence-based verdicts with auditability. It replaces manual review when you need structured scoring, transparent justification, and multi-agent debate across complex evaluation criteria.