Skip to content
Tools · Open-source AI reliabilityOpen source · github.com/hermes-labs-ai

Open-source reliability, in working code.

Free, open-source tools make our methods inspectable and give teams usable building blocks. Some are ready-to-install software; others are public reference implementations and research artifacts you can inspect, fork, or extend. Hire Hermes Labs when you need these methods diagnosed, integrated, and hardened in a specific production system.

See also: upstream contributions.

Tool
Purpose
Install · Source
§ Flagship software04 verified paths
lintlang
Static analysis for AI agent configs, tool descriptions, and system prompts. Zero LLM calls. Runs in CI.
pip install lintlang==0.3.1
github →
little-canary
Input-side prompt-injection detection via sacrificial canary-model probes.
pip install little-canary==0.3.3
github →
hermeneutic
Mines correction triples from chat logs and gates the next response before the same drift ships twice.
pip install hermeneutic==0.1.7
github →
agent-gorgon
Runtime policy guard for autonomous agents, with deterministic decisions and forensic evidence.
pip install agent-gorgon==0.1.6
github →
§ Public references02 source paths
fidelis
Fidelity-preserving agent memory with no LLM in the default retrieval path. The PyPI package named fidelis is unrelated.
git clone --branch v0.0.91 --depth 1 https://github.com/hermes-labs-ai/fidelis.git
python3 -m pip install ./fidelis
github →
hermes-rubric
Evidence-first structured scoring for advisory LLM evaluation; not a binary release gate.
pip install hermes-rubric==1.0.0
github →

We also keep reference implementations and research concepts public so others can inspect, fork, or develop them. Public does not mean flagship or production-ready. Browse the full GitHub catalog