terminal_labs
Research

Privacy Policy

Last Updated: Dec 2025

The "TL;DR" for Engineers

At Terminal Labs, we are building a tool for velocity, not a surveillance network. Here is the plain-English summary of how we treat your data:

  1. We do NOT train on your private code. The code generated on your machine is yours. It stays local. We do not upload your project files to our servers.
  2. We do NOT sell your data. We are an engineering company, not an ad network.
  3. We DO collect telemetry. We track which commands you run (e.g., npx snapapp -> Next.js RAG) and crash reports. This helps us fix bugs and prioritize features.
  4. We DO process your prompts. When you send a prompt to the CLI, it is sent to our inference server to generate the code. We do not use these inputs to train the base model.

The Full Policy

1. Information We Collect

  • Usage Data (Telemetry): We collect anonymous usage statistics, such as the framework selected (e.g., Next.js vs Python), the features requested (e.g., Pinecone vs Supabase), and the duration of the scaffold process.
  • Error Logs: If the CLI crashes, we receive a stack trace to help us identify compatibility issues across OS environments.
  • Prompt Inputs: Text prompts provided to the CLI are transmitted to our API for processing.

2. How We Use Information

  • Service Provision: To generate the requested application code and return it to your terminal.
  • Infrastructure Stress-Testing: As stated in our "Early Access" terms, we use telemetry to identify bottlenecks in our generation pipeline.
  • Model Evaluation: We may analyze aggregated prompt patterns (e.g., "Users are asking for LangGraph more often") to guide our internal dataset curation.

3. The "No-Training" Guarantee

Terminal Labs retrains its models (Vincent) weekly. This training is performed exclusively using:

  • Publicly available documentation (SDKs, API references).
  • Permissive open-source repositories (MIT/Apache licenses).
  • Our proprietary dataset of verified architectural patterns.

We do not use your private application logic, proprietary code, or local environment variables to train our models.

4. Third-Party Services

We use trusted third-party infrastructure to host our services (e.g., AWS, or similar providers). These providers process data solely on our behalf and are bound by strict confidentiality agreements.

5. Changes to This Policy

We treat our Privacy Policy like code. If we push a breaking change (a material update to how we handle data), we will notify you via the CLI or the email associated with your Early Access account.

6. Contact Us

If you have questions about this policy or want to request the deletion of your telemetry data, contact us at: privacy@terminalhq.com