Wanju vs. cloud agents
(Devin, Jules & co.)

Cloud agents clone your repository into their virtual machines, work autonomously, and hand back a pull request. That scales, and for well-specified tasks on code you're comfortable shipping to a vendor's cloud, it works.

Wanju makes the opposite trade, deliberately. Your repository is never cloned to Wanju's infrastructure. Agents run locally. The built-in agent uses your own OpenAI, Anthropic, or Gemini key and queues edits for approval. API keys are encrypted with your operating-system keychain. Credential-like files are unreadable to agents, search, and AI context, in every direction.

The control model differs too: Wanju's built-in actions are deny-by-default. Commands run through Wanju use an allowlist (tests, builds, lint), audited actions are recorded, and there's an emergency lock. After you approve a built-in proposal, your tests run automatically.

Cost: cloud agents bill per task or per seat. Wanju is free to use — you pay only your AI provider directly. See the full comparison table, or scan your repos to see what's worth finishing.