Shai CLI
A terminal coding agent reads project rules, works interactively or in scripts and can suggest fixes for failed shell commands.
- 630 ★ on the repository
- Last commit
- Apache-2.0
Broad access
Why this level
- The agent executes commands and project changes; HTTP mode exposes those actions through a service.
Install
Manual install
curl -fsSL https://raw.githubusercontent.com/ovh/shai/main/install.sh | shInstalls the stable Shai CLI. After installation, select a provider using `shai auth`.
Access and checks
How we reviewBroad accessThe tool can change data, run commands or handle secrets.
- The agent executes commands and project changes; HTTP mode exposes those actions through a service.
This is third-party code. Review the repository files before installing.
What it does
Shai runs development tasks in a terminal. Interactive mode supports coding conversations; scripted use accepts prompts as arguments or standard input. Saved sessions can be resumed by identifier. AGENTS.md supplies project context, including rules in nested directories. Custom agents select providers, tools and instructions. A separate shell-assistant mode watches failed commands and suggests fixes, while HTTP mode supports integration with a user-managed service.
Who it is for. For developers seeking a terminal agent with project rules and several execution modes.
Good fit when
- Pass a task from a script.
- Resume a saved coding session.
- Use the project’s AGENTS.md rules.
- Get a suggestion after a shell command fails.
Not a fit when
- Agent access to project files and tools is prohibited.
- You require operation without a model provider.
- The HTTP service would be exposed publicly without access restrictions.
Example request
Read AGENTS.md and propose a Shai plan for fixing the error on a test branch before edits.Limitations
The default is rate-limited anonymous OVHcloud access; select your provider with shai auth before normal use. The chosen model receives submitted context, so do not include private source without authorization. Scripted mode runs tasks with the user’s tool access; begin on a separate branch and inspect results. Shell-assistant mode installs hooks and observes terminal output, so enable it only where project rules permit.
How to disable. Run `shai off` if shell monitoring is enabled, then stop the agent and HTTP server. Inspect edits and saved sessions separately.
FAQ
How do I open interactive mode?
Run `shai`.
How do I resume the latest session?
Use `shai session latest` or `shai --latest`.
Where are custom agents stored?
In `~/.config/shai/agents/`.
How do I inspect MCP connections?
Use `/mcp` in interactive mode.
Which context file has priority?
AGENTS.md; SHAI.md remains a deprecated supplement.
Discussion
No questions or reviews yet.
Related
CLI toolsA terminal agent explores projects, prepares plans and makes edits or runs commands under a chosen approval policy.
An open source coding agent for the terminal and desktop with build and plan modes
DevChain runs a team of AI coding agents on your own machine with a shared board, chat, code review and live terminals.
An open coding agent for the terminal, browser and desktop that works with models from OpenAI and Claude to Ollama
A local control panel for coding agents: it runs Claude Code and Codex, keeps a task queue and merges verified work into main.
A terminal coding agent explores projects, edits files and runs checks using your chosen model and command policies
README in short
Shai supports separate agent configurations and on-demand skills, allowing tasks to use different providers and tool sets.