wrokin puts an AI engineering team inside your repos. Install once, and
wrokin's agents react to your PRs and issues and do real work — on the model
you choose, with your API key.

Under the hood, wrokin runs on the Pi agent, with our own custom workflows,
context-gathering (Scout), and additional harnesses and guardrails layered on top
to make it more effective and safer.

What the team does today

• Code Review — diff-aware review that catches bugs, smells, and risky changes.
• Security Audit — flags injection, secrets, and unsafe patterns on every PR.
• Issue Triage — labels, summarizes, and routes new issues automatically.
• Project Management — keeps issues and PRs organized and moving.
• Scout — a zero-token context pass that finds the right files and your
  AGENT.md / CLAUDE.md guidance before any agent runs.

Coming soon

• Builder (autonomous) — turns an issue into a working pull request. In beta;
  rolling out to paid plans after launch.
• Pro, Team & Business plans — more volume, recipes, and (on Business) opt-in
  auto-merge. Available once paid plans go live.

Bring Your Own Key (BYOK)
Connect your own key from any of 6 providers — Anthropic, OpenAI, Google, Mistral,
DeepSeek, OpenRouter — and assign a model per role. You pay your provider directly
for tokens; wrokin never marks up inference.

Privacy & honesty
wrokin processes your code to do its work and sends the relevant context to the
LLM provider you selected, using your key. Keys are encrypted at rest, scoped
per installation, and never logged. wrokin is not a substitute for human review —
agents can be wrong; you stay in control of what merges. We don't claim
end-to-end encryption, "we never see your code," or any certification we don't hold.