New APINew API
AstrLink · On-device privacy gateway

Secrets stay local. Agents keep working.

Secrets, phone numbers and names are swapped for placeholders on your machine, then restored when the reply arrives.

Visit astrlink.comStar on GitHub

Open source · macOS / Windows / Linux · New API ecosystem

Your computer
Your agentClaude Code · Codex
AstrLinkRedact · restore
New APINew APIKeys · quotas · routing
Upstream modelsOpenAI · Claude · Gemini
What your agent reads
# .envDB_URL=postgres://admin:q7Vt2LmX9pRk@10.0.3.17/prodONCALL=Zhang Wei <zhangwei@example.com>
What the model receives
# .envDB_URL=postgres://admin:secret@IP/prodONCALL=name <email>
Replies are restored on your machine, so your agent carries on.Fictional example. Results depend on your policies.
Why it matters

Agents that do more can leak more

It reads files on its own

Agents open .env files and logs. What you never pasted can still get sent.

Every hop sees plain text

Provider APIs, third-party services and team gateways all see the full request.

Customer data lands in logs

Names, phone numbers and internal IPs can stay in a provider's request logs.

Detection

Two detectors, both on your machine

Built in

Built-in rules

Emails, phone numbers, card numbers and common key formats, right after install.

Local model

AstrLink Guard

Names, addresses, account IDs and credentials buried in code and logs.

~35 MB · runs offline · Chinese and English

You choose what happens on a match

Warn and continueBlock the requestRedact and continue

Placeholders in replies are restored on your machine.

If the privacy model fails to load, requests are rejected.

With New API

Connect New API in three steps

New API handles quotas. AstrLink handles privacy.

  1. 1

    Create a token in New API

    Give AstrLink its own token, with model and quota limits.

  2. 2

    Add it as a provider

    Enter your New API URL and token in AstrLink, then fetch models.

  3. 3

    Point your agent here

    Use the local address as the base URL and an AstrLink access token as the key.

Agent settings
Base URL
http://127.0.0.1:8317/v1
API Key
AstrLink access token
Model
A model ID from New API

If the port is taken, AstrLink picks another. Use the one shown in the app.

What else it does

Manage your AI tools in one place

Subscriptions and APIs

Codex, Claude and Grok subscriptions, provider APIs and New API.

Routing and failover

Model redirects, provider priority and automatic retries.

Multiple protocols

OpenAI, Anthropic and Gemini endpoints with protocol conversion.

Request records

Traces, errors and estimated cost for every request.

FAQ

How does it relate to New API?

Both are part of the New API ecosystem. New API manages keys and quotas on the server, AstrLink handles sensitive data on your machine, and you can chain them.

Is my content still sent to AI providers?

Yes. Your chosen providers still run inference, on content your policies have processed.

Do I need the privacy model?

Built-in rules work right away. Download the model in the app when you need to catch names and addresses.

Does it guarantee nothing leaks?

No. Detection can miss things and flag false matches. Try a dry run first and allowlist false matches.

Hand off the work. Keep the secrets.

Free and open source. Still early, so feedback is welcome.

Visit astrlink.comStar on GitHub