Obscurify.ai provider for OpenClaw

An OpenClaw plugin that adds Obscurify.ai as a model provider. After you install it, Obscurify.ai shows up in onboarding and in the model picker, and models are addressed as obscurify/<lab>/<model>, for example obscurify/deepseek/deepseek-v4-flash.

Obscurify.ai is a privacy-focused, OpenAI-compatible API to 200+ models (GPT, Claude, Gemini, DeepSeek, GLM, Kimi, Qwen, Llama and more). Text requests only use zero-data-retention routes, prompts and responses aren't stored on its servers, and accounts are a random account number with no email.

Install

openclaw plugins install clawhub:@obscurify/openclaw-provider
# or from npm
openclaw plugins install npm:@obscurify/openclaw-provider

Needs OpenClaw 2026.10.8 or newer.

Get an API key

1. Create an account at obscurify.ai. You get a random account number: no email, no password.
2. Add a balance. The API works on every plan with a positive balance. Frontier models need the Casual plan or higher, and uncensored models need Pro.
3. On obscurify.ai/account, click Show API Key and copy the sk-obs1_... key.

Configure

Choose Obscurify.ai in openclaw onboard, or pass the key directly:

openclaw onboard --auth-choice obscurify-api-key --obscurify-api-key sk-obs1_...

The plugin also reads OBSCURIFY_API_KEY from the environment.

openclaw models list --all --provider obscurify --refresh
openclaw models set obscurify/deepseek/deepseek-v4-flash
openclaw agent --local --model obscurify/z-ai/glm-5.3 --thinking low --message "Hello"

What it does

• Model list: comes live from Obscurify.ai's public GET https://obscurify.ai/v1/models (no key sent). Decision models, which only answer structured questions, are left out. Five models ship as a static fallback with prices: DeepSeek V4 Flash (default), GLM 5.3, Kimi K2 0905, Qwen3 Coder 480B and Llama 3.3 70B.
• Chat: goes through OpenClaw's OpenAI-compatible transport to https://obscurify.ai/v1, with streaming, tool calls, reasoning (--thinking) and token usage in streams.
• What it doesn't do: no background processes, no telemetry, no file access. Requests cost money from your Obscurify.ai balance at each model's price (obscurify.ai/model_cards).

Tested

With OpenClaw 2026.10.8 on Linux (2026-10-28), using a Pay As You Go key, via scripts/e2e.sh:

• installed the packed plugin
• listed 303 models
• openclaw agent --local answered on Llama 3.3 70B
• on DeepSeek V4 Flash with --thinking low, the agent wrote a file through its tools

License

MIT