PolyBrain: multi-agent tool for Hermes Agent

Multi-model orchestration skill for Hermes Agent with parallel subagents and citation checks

What PolyBrain does

An orchestrator role decomposes the objective into a JSON task plan. Researcher roles use web and browser toolsets and must include URLs, builder roles use terminal and file toolsets, a synthesizer merges the outputs, and a verifier checks each claim against its cited source and returns PASS or FAIL. Uncited claims are dropped. Each role can be pointed at a different model or provider in config.yaml, with a fallback model for transient failures.

According to the README, this differs from Hermes's native delegate_task and Kanban, which start subagents bound to the same model config. PolyBrain also recovers orchestrator JSON when models add extra prose, retries transient failures, and saves every run in a timestamped artifact folder. Usage is conversational: tell Hermes to use PolyBrain, or pipe an objective into scripts/orchestrate.py. A validate_config.py script checks the configuration. The README says it was inspired by the orchestration pattern behind Perplexity Computer.

Key features

Role-based routing of orchestrator, researcher, builder, synthesizer and verifier to separate models
Parallel execution of research and build subtasks
Citation enforcement that drops uncited claims
Claim-by-claim verification with PASS or FAIL results
Retry logic and timestamped artifact logging for every run

When to use it

Research a company and its competitors with sourced, verified claims
Use a cheaper model for research and a stronger one for synthesis
Audit what each role produced from the saved run artifacts

Who it is for: Hermes Agent users who want different LLMs handling different roles in one sourced research or build workflow.

How it fits with Hermes Agent

Built for Hermes Agent as a skill installed under ~/.hermes/skills/research/polybrain and invoked from a Hermes chat.

How to install PolyBrain

These commands are copied from the project's README. Check the repository for the latest steps before you run them.

git clone --depth=1 https://github.com/mosesman831/PolyBrain.git /tmp/polybrain
rm -rf /tmp/polybrain/.git
cp -r /tmp/polybrain ~/.hermes/skills/research/polybrain
rm -rf /tmp/polybrain

FAQ

What is PolyBrain?

PolyBrain is a Hermes Agent skill that decomposes an objective, runs specialized roles in parallel on models you choose, and verifies every claim against cited sources before returning an answer.

How do I install PolyBrain?

Clone the repository, remove its .git folder, and copy it to ~/.hermes/skills/research/polybrain. Then edit config.yaml with your model aliases and run scripts/validate_config.py.

Is PolyBrain free and open source?

Yes. The repository is published under the GPL-3.0 license.