More screenshots
Works with the AI Agents you already use
Give Your Agent a Local Brain (Ollama)
Run your agent's thinking on your own computer with Ollama — no API key, no per-call cost, nothing leaves the machine.
$9
Give Your Agent a Local Brain (Ollama)
Example session with this skill installed
Configure my agent to use a 20B model on Ollama with a fallback to GPT-4o only if the local server is down and an API key is present.
- Read your context and instructions
- Compiled the give your agent
I have configured the agent brain to prioritize local inference using the `gpt-oss:20b` model via Ollama. The connection logic now includes a health-check step and a conditional fallback routine. ## Local-First Brain Configuration ```python import requests import openai # Configuration for Local-First Logic LOCAL_URL = "http://localhost:11434/v1" LOCAL_MODEL = "gpt-oss:20b" FALLBACK_MODEL = "gpt-4o" HOSTED_KEY = "sk-..." # Should be pulled from env def get_chat_completion(messages, sensitive=False): # 1. Health-check Ollama try: response = requests.get("http://localhost:11434/api/version", timeout=2) response.raise_for_status() server_live = True except Exception: server_live = False # 2. Execution Logic if server_live: client = openai.OpenAI(base_url=LOCAL_URL, api_key="ollama") return client.chat.completions.create(model=LOCAL_MODEL, messages=messages) # 3. Fallback logic if not sensitive and HOSTED_KEY: client = openai.OpenAI(api_key=HOSTED_KEY) return client.chat.completions.create(model=FALLBACK_MODEL, messages=messages) # 4. Safety catch raise ConnectionError("Local brain is down. Task queued to protect privacy.") ``` **Next steps** 1. Run `ollama pull gpt-oss:20b` in your terminal to download the 12GB model weights. 2. Verify the server is responding by running the smoke test in `examples/brain-wiring.py`. 3. Set your `keep_alive` parameter in Ollama to prevent model unloading between agent tasks.
Connects securely to your tools. The creator never sees your data.
About this skill
A personal agent reads your money, your messages, and your plans. Sending all of that to a hosted model by default is a choice, not a requirement. This skill wires the agent's brain to a local model through Ollama — a free program that runs models on your own machine and answers through the same kind of endpoint the hosted APIs use. Inside the package: the skill file covering install, model choice, wiring, and the fallback rules; a hardware table that matches model size to your machine (roughly: 8 GB of RAM runs a small model, 16 GB runs a 20B-class model well, 32 GB and up runs 30B-class); and a short Python smoke-test file that proves the wiring works before you debug anything else. A run goes like this: install Ollama, pull the model the table picks for your hardware, and point your agent at http://localhost:11434/v1 with the placeholder key the local setup expects. Before each session, the agent health-checks the local server. If the local brain is down, it steps down to a smaller local model first, and uses a hosted model only if you set one up yourself. Work involving private data waits for the local brain instead of being sent out.
Find more skills like this on the Claude Skills Marketplace, the Codex Skills Marketplace and the Gemini Skills Marketplace, or browse All Claude Skills.
Requires manual installation of Ollama. 20B+ models need ~16GB+ RAM for fluid performance. Port 11434 is unauthenticated; do not expose to public networks.
How to use
- 1
Connect your agent to the Agensi MCP
Copy the prompt and paste it into Claude, ChatGPT, Cursor or any agent that supports MCP. It sets up the connection for you. Manual setup
- 2
Sign in to Agensi
Your agent shows a sign-in link. Sign in and click Allow.
- 3
Ask for what you need
Your agent loads this skill and uses it on your task. It always asks before buying a skill.
Reviews
No reviews yet
Be one of the first to try it. Every listed skill goes through the trust checks below.
Security scanned
Checked by our 8-point scan before listing
Fresh listing
Recently published to Agensi
Money-back guarantee
Doesn't do what its page promises? You get your money back.
Trust & safety
Security scan
Security scan passed 6 days ago
- Software
- Ollama installedA local model that fits your RAM/VRAM
Python 3.10+
- Optional
- OpenAI API key for hosted fallback