DevDiff

Appearance

Privacy Architecture & Data Protection ​

Privacy is the foundational guarantee of DevDiff. Your source code represents your core intellectual property, customer data, and competitive advantage. DevDiff is engineered so that your code never leaves your local workstation unless you explicitly configure a third-party AI provider.

Privacy Guarantees ​

5 Privacy Pillars ​

1. Local-First Storage & Memory Indexing ​

  • Codebase memory indexes (.devdiff/memory/codebase-index.json) are generated and stored strictly inside your local repository folder.
  • Memory indexes are stored in human-readable, plain-text JSON format, allowing full transparency and auditability.

2. Bring Your Own AI (BYOAI) Architecture ​

  • You Control the Model: DevDiff supports local Ollama models (llama3.2, codellama, deepseek-coder, qwen2.5-coder) and browser WebGPU execution.
  • Direct API Credentials: When cloud providers (OpenAI, Anthropic, Gemini) are used, your API keys interact directly with provider endpoints. DevDiff operates no proxy servers, middleman services, or data collection hubs.

3. Zero Data Retention & Zero Training ​

  • Local execution leaves 0 external data footprint.
  • When cloud provider endpoints are used, DevDiff formats requests to comply with Zero Data Retention (ZDR) enterprise policies. Your code is never used to train foundation models.

4. Automatic Secret Redaction ​

  • All code snippets, commit diffs, and AST entities pass through RedactionEngineV2 prior to processing.
  • API keys, credentials, private keys, and passwords are automatically masked with replacement tokens (e.g. [REDACTED:API-Key]).

5. Compliance Alignment ​

  • GDPR & CCPA Compliant: Zero personal data collection, zero third-party tracking, and instant local data deletion (devdiff memory clear).
  • SOC 2 & ISO 27001 Ready: Meets strict enterprise data boundary requirements.