ollama-paperclip-adapter

Paperclip external adapter for Ollama — run local LLMs as Paperclip agents without any API key or cloud dependency.

Setup

1. Install Ollama

Download from ollama.com/download or:

curl -fsSL https://ollama.com/install.sh | sh

2. Pull a model

ollama pull llama3.2
# or any other model:
ollama pull qwen2.5-coder
ollama pull mistral
ollama pull deepseek-r1

3. Configure the adapter in Paperclip

Go to Settings → Adapters → Add adapter and set:

adapterType: ollama_local
adapterConfig:
  model: llama3.2               # optional if at least one model is installed
  # Optional:
  baseUrl: http://localhost:11434   # default
  maxTokens: 8192                   # default
  temperature: 0.7                  # default
  timeoutMs: 120000                 # default
  customSystemPrompt: ""            # append extra instructions

No API key is required. Ollama runs fully locally. If model is omitted, adapter uses the first available model from ollama /api/tags.

Supported models

Any model pulled into your local Ollama. Popular choices:

| Model ID | Description | |---|---| | llama3.2 | Meta Llama 3.2 | | llama3.1 | Meta Llama 3.1 | | qwen2.5-coder | Alibaba Qwen 2.5 Coder | | qwen2.5 | Alibaba Qwen 2.5 | | mistral | Mistral 7B | | gemma3 | Google Gemma 3 | | phi4 | Microsoft Phi-4 | | deepseek-r1 | DeepSeek R1 (reasoning) | | codellama | Meta Code Llama |

Remote Ollama

To use a remote Ollama server (e.g. a GPU machine on your local network):

adapterConfig:
  model: llama3.2
  baseUrl: http://192.168.1.100:11434

Cost

Ollama models run locally — costUsd is always 0.

Development

npm install
npm run build
npm test