Crowe Logic

The CroweLM agent console for the terminal. One command, a real tool loop, and a model chain that falls forward when a provider stalls.

Crowe Logic runs as an interactive console or a single prompt. It carries its own tool registry, reads and edits files, runs shell commands under an autonomy policy you set, searches the web, and streams reasoning as it works. Tiers are selectable by effort rather than by model name, so /effort max asks for more thinking without you tracking deployments.

Proprietary software. See LICENSE. Free tier available with no card and no keys.

Install

pip install crowe-logic

npm i @michaelcrowe11/crowe-logic

Setup

1. Deploy the CroweLM-backed Foundry models on Azure AI Foundry
2. Configure credentials:

cp .env.example .env
# Fill in the Azure endpoints and API keys for the CroweLM resources

3. Authenticate with Azure:

az login

4. Verify the model stack:

crowe-logic deploy

5. Start chatting:

crowe-logic

Usage

crowe-logic              # Interactive chat (default)
crowe-logic chat         # Interactive chat session
crowe-logic run "prompt" # Single prompt, get response
crowe-logic deploy       # Verify provider health across the CroweLM stack
crowe-logic models sync --account <account> --resource-group <resource-group>
crowe-logic status       # Show agent status
crowe-logic tools        # List available tools

CroweLM Apex is wired to stream reasoning summaries through both the terminal UI and crowe-logic headless when the Azure Responses API emits them, so hosts that consume the JSON event stream can render reasoning deltas before answer tokens.

Extra Models

The base MODEL_CHAIN can now be extended without editing source. Crowe Logic will load extra model entries from:

• CROWE_LOGIC_EXTRA_MODELS_JSON
• CROWE_LOGIC_EXTRA_MODELS_PATH
• config/models.extra.json
• ~/.config/crowe-logic/models.extra.json
• ~/.crowe-logic/models.extra.json

Start from config/models.extra.example.json and save your generated file as config/models.extra.json, or point CROWE_LOGIC_EXTRA_MODELS_PATH at it.

If you already have Azure deployments and want to turn them into Foundry model entries, generate the file with:

crowe-logic models sync --account <account> --resource-group <resource-group>

By default that writes to ~/.config/crowe-logic/models.extra.json, which the runtime already loads. Use --output config/models.extra.json if you want a project-local registry instead.

You can also sync from a saved Azure CLI payload:

az cognitiveservices account deployment list --name <account> --resource-group <resource-group> --output json > deployments.json
crowe-logic models sync --input deployments.json --output config/models.extra.json

Tools (79)

MCP Server

The Crowe Logic platform is also available as a standalone MCP server for any MCP client:

uvx crowe-logic-mcp

See crowe-logic-mcp for configuration details.

Specialized Agents

Environment Variables

Docker

docker pull michaelcrowe1111/crowe-logic:latest
docker run -it --env-file .env michaelcrowe1111/crowe-logic

GPU variant for fine-tuning:

docker build --target gpu -t crowe-logic:gpu .

Requirements

• Python 3.10+
• For self-hosted tiers: an Azure AI Foundry account with the CroweLM lineup deployed. The deployment matrix is provided with onboarding.
• Azure CLI (az login) for authentication

Author

Michael Crowe -- Crowe Logic, Inc.

License

Proprietary. Copyright (c) 2026 Crowe Logic, Inc. All rights reserved. See the LICENSE file for terms. Licensed, not sold: use requires the no-cost tier or a current paid plan, and redistribution is not permitted.

Releases before 0.7.1 shipped an Apache License 2.0 text and were distributed under it; this change is not retroactive.

Licensing questions: michael@crowelogic.com