Frequently asked questions

Common questions about Sleev.

How does Sleev work?

Sleev runs as a local proxy between your coding tool and your model provider. Your tool still talks to Anthropic, OpenAI, Codex, Kimi, or another configured upstream, but requests pass through Sleev first. Sleev intelligently compresses context that is deemed stale or no longer needed for the session, focusing on low-signal noise rather than important working context. Cache hit rate is treated as a key optimization dimension, so Sleev avoids rewriting or removing context in ways that would break provider caching unless the expected savings justify it.

Can Sleev run in limited-internet environments?

Yes, for approved accounts. Offline licensing lets the local gateway verify a signed license on the machine instead of relying on normal account sign-in, regular online account checks, or usage-pattern uploads to Sleev. The model provider itself may still require network access from that environment. See the offline licensing docs for the activation and renewal flow.

Do you store my conversations?

Sleev does not send your conversation history, prompts, code, or model responses to our servers. The local gateway sends usage reports needed for the dashboard and billing, and the server replies with whether the account still has usage available. Those reports are limited to account, token, session, and usage metadata, not the content of what you are doing with your coding agent.

Do you collect provider credentials?

No. Your provider credential remains with your coding tool or provider configuration. The tool sends it to Sleev with each proxied request, and Sleev forwards it upstream unchanged without storing it.

How do I stop using Sleev?

Remove the Sleev-specific entries from your coding tool's config file and it will go back to direct provider calls. The Deconfigure section shows what to remove. This stops the harness from using Sleev, but a local gateway process may still be running. To stop and remove the gateway too, open the sleev TUI and sign out. This will also deconfigure any detected coding harness' config file.

Which providers and tools are supported?

Sleev can route requests to Anthropic, OpenAI, Codex, Kimi Coding, Moonshot AI, and opencode-go, depending on the coding tool you configure. Claude Code, OpenCode, and Codex CLI are documented in the configuration docs.

How can I get in contact or give feedback?

Email support@sleev.ai or join the Sleev Discord to ask questions, report issues, or share feedback with the team. Every way to reach us is listed on the contact page.

Is there a service level agreement (SLA)?

During early access, we do not offer a formal SLA. Uptime commitments may come with a future enterprise tier.