# Codag > Codag cuts coding-agent costs without changing how developers work. One setup command attaches it to Claude Code and Codex; it teaches the agent to read large tool results efficiently, keeps exact content retrievable on the developer's machine, and passes everything else through unchanged. Codag Inc. is based in San Francisco and backed by Y Combinator. ## How it works - Install: `curl -fsSL https://codag.ai/install.sh | sh`, then run `codag setup`. There is no wrapper command; developers keep using their agents normally. - Codag observes each tool call, classifies it, and reduces large eligible results (logs, test/build/lint output, search results, file listings, document reads, agent handoffs) to the evidence the agent needs. - Source code, diffs, configuration, and unrecognized output always pass through unchanged. Failures of any kind mean byte-for-byte passthrough. - Exact omitted content stays encrypted on the developer's machine (7 days, 1 GiB cap) and is retrieved locally by line range, JSON path, or group. - Model traffic goes from a local proxy directly to the developer's provider. Codag's cloud never retains raw content; billing runs on contentless metrics. ## Pricing Codag is sold to teams; there is no self-serve individual tier. Pilots (14 days, no card, on the team's real workload) are arranged on a demo call. - Team: $499/month, 200 GB of eligible tool output included, $0.50 per additional GB, up to 25 seats. - Enterprise: custom contractual volume, SSO/SAML, VPC peering, version pinning, custom SLAs. - Caps mean passthrough, never a broken session. Dollar savings shown in the product are estimates from public model prices. ## Docs - [Quickstart](https://codag.ai/docs): install, codag setup, verify. - [Supported agents](https://codag.ai/docs/agent-setup): Claude Code and Codex; reversible attach. - [CLI reference](https://codag.ai/docs/cli): setup, status, savings, doctor, disable, uninstall, auth, config. - [Privacy and data flow](https://codag.ai/docs/privacy-data-flow): what leaves the machine. - [REST API](https://codag.ai/docs/rest-api): POST /v1/actions/reduce plus usage, model-price, and policy endpoints. - [SDKs](https://codag.ai/docs/sdks): Python (`pip install codag`), TypeScript (`npm install @codag/sdk`), Go. ## Links - [Overview](https://codag.ai/) - [Pricing](https://codag.ai/pricing) - [Open Source](https://codag.ai/open-source): codag-sdk, codag-drain, codag-visualizer; the production runtime is a signed closed-source binary with verifiable releases. - [Vision](https://codag.ai/vision) - [Blog](https://codag.ai/blog) - [Support](https://codag.ai/support) - [Privacy](https://codag.ai/privacy) - [Terms](https://codag.ai/terms)