AI agent — Eve (champrep ai / eve)
eve (or champrep ai / champrep eve) launches Eve, CHAMPREP’s intelligent AI pair-programming and ecosystem agent in your terminal. Eve knows both names:
“I am Eve. I’m also sometimes referred to as champrep ai, and this is what I do on this platform: assisting with coding, workspace exploration, multi-agent parallel workflows, and running development tasks.”
Eve is the same CHAMPREP AI you use across the web, desktop, and mobile apps, with superpower capabilities in the terminal: reading, editing, and searching local code, managing background processes and shells, running multiple subagents in parallel, and securely accessing your cloud CHAMPREP services.
What Eve can do
Section titled “What Eve can do”- Intelligent pair-programming on local code. Read, search, and edit files in the current directory (and any directory added via
--add-dir), run tests, and usegit. - Startup logo animation. When Eve starts up, it displays an animated glowing ChamPrep logo with the official brand colors. Replay it at any time with
/logo. - Live background processes and shells tracking. Monitor all active background shells, compilers, dev servers, PIDs, and ports right on your statusline (
🐚 N shells) or run/shells//procsto inspect their output. - Massive multi-agent parallelism. Split complex missions into parallel subagent tasks (up to 32 concurrent agents) that execute simultaneously, keeping the main context clean.
- Multi-hour execution resilience. Built for complex development workflows and deep refactors that can last hours, with state-suspension tolerance supporting up to 2,000 turn cycles.
- Use your CHAMPREP services. Search Drive, compose Mail, manage Work tickets, check Calendar, and invoke connected MCP servers.
- Stay in complete control. Every file edit and command execution requests permission unless a permission mode or rule allows it. Secrets and sensitive files are protected.
Before you start
Section titled “Before you start”- Install the CHAMPREP CLI. The installation provides both
champrepand the standaloneeveexecutable. - Confirm that your plan includes the CLI agent. See Plans and limits.
Sign in
Section titled “Sign in”For a person at a terminal, use browser sign-in:
champrep auth loginA CLI browser sign-in can use the agent’s local tools automatically.
For automation, store an API key that includes the ai:agent scope:
champrep keys create --name "ci-agent" --scope ai:agentchamprep auth tokenai:agent must be selected by name. A key that has only broader scopes cannot
run local tools. See API keys for CLI automation for handling
keys safely, and Headless and CI usage for pipelines.
Start Eve
Section titled “Start Eve”Open a terminal in your project and run either:
cd ~/src/my-projecteveOr using the champrep binary:
champrep ai# or:champrep eveBoth commands launch the exact same interactive agent session.
Type a request and press Enter. Type / to see every command, or /help
for commands and key bindings. The status line displays Eve’s active profile, model,
context usage, git branch, permission mode, and active background shells (🐚 N shells).
Useful start options (works with either eve or champrep ai):
| Option | Effect |
|---|---|
eve --continue (or champrep ai --continue) | Reopen the latest conversation started in this directory. |
eve --resume (or champrep ai --resume) | Pick any conversation to reopen, or pass an ID. |
eve --model NAME | Use a specific model for this session. See Models. |
eve --permission-mode MODE | Start in ask, accept-edits, or plan mode. |
eve --add-dir PATH | Let the agent read and write another directory too. Repeatable. |
eve --worktree [NAME] | Work in a new git worktree. See Worktrees. |
eve --agent NAME | Start the session as one of your custom agents. |
eve --keep-checkpoints | Keep file checkpoints after the session ends. |
Run one prompt and exit (Headless mode)
Section titled “Run one prompt and exit (Headless mode)”Use -p with either eve or champrep ai to run a single prompt without the interactive TUI:
eve -p "summarize what this repository does"# or:champrep ai -p "check for breaking changes between main and this branch"Headless runs are read-only unless you allow more. See Headless and CI usage.
Conversations follow you
Section titled “Conversations follow you”Conversations are saved to your CHAMPREP AI account, not to your machine.
/resumein the agent lists conversations from every surface, including the web, desktop, mobile, and other terminals.- Conversations you start with Eve also appear in your CHAMPREP AI conversation history on the web. Continuing one there cannot touch your machine, because local tools only run in the terminal.
/continueand--continuereopen the latest conversation for the current directory.