What is HammerAI?
HammerAI is a local-first AI character chat and writing application with an external LLM Proxy Mode.
This HammerAI guide uses the official Proxy Mode fields: an API base ending in /v1, exact provider model name, API key, and an optional context length that should only be set when known. Proxy Mode routes character-chat model requests to an external API while HammerAI continues to provide the local conversation interface. Use the API root and exact model name, leave context length empty unless the verified value is known, and reload the chat before testing a short response and a longer session.
HammerAI's official Proxy Mode sends Character Chat generation requests to an external OpenAI-compatible service. It uses four Proxy fields and does not require a local model download or a HammerAI hosted-model account.
Enable Proxy from the official HammerAI interface
- Open any Character Chat.
- Switch the right-hand panel to Settings.
- Find the LLM menu.
- Scroll to the bottom and choose Proxy.
- Complete the Proxy form with the values below.
Map the four fields to TokenHub
- Proxy URL:
https://us-api.tokenhub.com/v1 - Proxy Model: the complete model ID from the TokenHub model catalog
- Proxy API Key: your TokenHub API key
- Proxy Context Length: optional; leave it empty when the real limit is unknown
The official Proxy examples use API bases such as LM Studio's /v1, not a complete request path. Recent HammerAI versions may trim an accidentally supplied /chat/completions or /completions suffix, but enter the /v1 root directly.
Account for the difference between Web and Desktop
TokenHub is a remote HTTPS service and normally needs no additional CORS setting. CORS instructions in HammerAI apply when the Web app connects to a local service such as LM Studio. Do not transfer those local-server instructions to the remote TokenHub URL.
Make the Proxy configuration take effect
Save, send one short message, and confirm its model ID in TokenHub request logs. HammerAI notes that some services require a full app reload before a connection is established. If all four fields are correct but no response arrives, reload HammerAI and test again.
For long-chat failures, clear or correct Proxy Context Length first. For 401, inspect Proxy API Key. For 404, confirm Proxy URL stops at /v1. If the model does not exist, copy Proxy Model again. Checking these four fields is more useful than applying unrelated local-model parameters.
Official reference
Choose a compatible HammerAI model
HammerAI Proxy Mode FAQ
What should I enter as Proxy URL?
Enter the TokenHub API base ending in /v1. HammerAI’s official Proxy Mode examples use a base URL rather than a complete request path.
Should I fill in Proxy Context Length?
It is optional. Leave it blank when you do not know the model’s maximum context; otherwise enter the accurate value.
Why does the proxy not respond after I save?
Check the Base URL, exact model name, and API key, then reload the chat. The official HammerAI guide notes that some services require a full reload.
Where does HammerAI send data in Proxy Mode?
Model requests for that character chat go to the configured external API while HammerAI provides the local interface. Review each service's data handling before use.
How do I test HammerAI Proxy Mode safely?
Reload the chat after saving, send a short message, and confirm the exact model and request status in TokenHub logs before a long conversation.
References
This setup is based on TokenHub and official documentation, last verified on 2026-09-08.
