specthread

Providers

Built-in and custom AI providers, with the API endpoint each one calls. Keys are saved to this browser's local storage — anyone with access to this device can read them.

ProviderAPI keyEndpoint

Models

The models offered in the picker under the chat message box. Tick one as the default.

DefDisplay nameModel idProvider

Model behavior

The system prompt and generation parameters used when calling the model.

System prompt

~106 tokens

Temperature

Higher is more creative, lower is more deterministic.

Max tokens

Cap on the response length. Raise it if long answers get cut off.

Chat

How the composer behaves.

Send on Enter

When off, Enter adds a newline and ⌘/Ctrl+Enter sends.

Default request headers

Attached to every proposed request.

Target API auth

Pre-filled onto proposed requests. The base URL is chosen from the spec bar.

Auth header name

Auth header value

Execution guards

Applied by the execute proxy once the backend is wired up.

Timeout (ms)

Max response (KB)

Reject private hosts

Block requests that resolve to private/loopback IPs (SSRF guard).

Endpoints

How the endpoint list is presented.

Group by tag

Show method filters

Show schema usage

Cross-reference each endpoint with the schemas it uses, and each schema with the endpoints that use it.

Conversations

Chats are saved per spec in this browser. Open the home page to pick one back up.

Save conversations

When off, chats live only for the current session and are never written to storage.

Conversations kept per spec

Starting a new conversation past this evicts the oldest. 1 means a new one replaces the previous.

Clear saved conversations

Permanently delete every saved conversation from this browser.

Data

Everything is stored in this browser's local storage.

Export settings

Copy the JSON, or download it as a file you can re-import later.

Import settings

Replace all settings from an exported JSON file.

Reset to defaults

Restore every setting to its original value.