Grimnir

Fast to start.

Built to go deep.

Grimnir helps operators move quickly without buying into heavyweight platform overhead. It is designed to be lightweight, flexible, transparent, and increasingly capable as better AI models become available over time.

Most AI security platforms push you toward more stack, more setup, more drag.

Grimnir pushes toward faster starts, deeper workflows, and clearer control.

Easy to start

The first run should feel like using a tool, not adopting a platform.

Lightweight by design

Less infrastructure, fewer moving parts, and less friction between intent and action.

Flexible workflows

Opinionated enough to help, adaptable enough to fit different operators and testing styles.

Transparent execution

Runs should be readable, reviewable, and useful after the fact — not just impressive while they happen.

Focused where modern security work is moving.

Network Pentesting

Reconnaissance, enumeration, exposed services, and the signal that drives the rest of an engagement.

Web Application Testing

API behavior, auth flows, business logic, and practical app-focused testing workflows.

AI / LLM Security

Prompt injection, unsafe tool use, retrieval issues, model misuse, and modern AI-targeted testing patterns.

Expanding Capabilities

Code Review

Bringing deep code analysis seamlessly adjacent to practical security scenarios.

Grimnir should grow more capable as the model layer improves.

Grimnir is model-flexible by design. It works with strong models available today and becomes more capable as better models arrive — without tying your workflow to a single provider forever.

Supported Today

Claude Integration

Harnesses Claude's deep reasoning and context window to triage massive recon outputs and write precise exploitation narratives for targets.

Expanding Access

Gemini Integration

Plugging directly into native multimodal inputs to broaden the scope of what can be evaluated in an engagement scenario.

Future Direction

Future Ecosystem Support

Built structurally to adapt to new APIs, advanced Codex-style execution layers, and whatever the frontier of language and reasoning models brings next.

Built to become more teachable over time.

Grimnir is being shaped to not only execute workflows, but also to make them easier to understand, review, and learn from. Over time, that means clearer explanations, better post-run review, and a more educational operator experience.

A focused assistant for people who want momentum, depth, and control.

Install & Go

# Clone and install requirements

> git clone https://github.com/DoctorGoz/grimnir
> cd grimnir && pip install -r requirements.txt

Open GitHub Repository