# agend

> agend provides persistent, isolated Linux environments that AI agents can operate through the Model Context Protocol (MCP). Its terminal tools support interactive programs, retained sessions, input-wait events, background tasks, file transfers, and public HTTPS previews. Bring your own agent and model access.

Last reviewed: 2026-10-09.

Start with the full reference for a concise product and setup guide, or the CLI README for commands and examples. The local `agend mcp` command provides the stdio MCP server; environment execution happens remotely. Humans can use `agend connect` for a live terminal and `agend watch` to observe an agent's interactive session.

`input_wait: true` reports an input-wait event observed during response collection. `status: awaiting_input` alone means the interactive process is alive. Neither establishes application health. Use the reference below for the exact semantics and an application's own health checks for service readiness.

Current prices, trial limits, and resource allocations are documented on the pricing page and in the account's available profiles. The installed server's tool schemas define accepted arguments.

## Documentation

- [Full agend reference](https://agend.sh/llms-full.txt): Markdown product overview, installation, MCP configuration, interactive terminal semantics, tools, lifecycle, and troubleshooting.
- [CLI README](https://raw.githubusercontent.com/agend-sh/cli/main/README.md): Maintained Markdown setup guide and CLI/MCP reference for Linux, macOS, and Windows.
- [CLI repository](https://github.com/agend-sh/cli): Public client source, issue tracker, and release history.
- [Latest CLI release](https://github.com/agend-sh/cli/releases/latest): Signed binaries and release notes.

## Product

- [Overview](https://agend.sh/): Persistent interactive terminals for AI agent workflows; setup, features, and frequently asked questions.
- [Terminal readiness](https://agend.sh/terminal-readiness): Input handling, process completion, and application health, with workflow examples and historical issue reports.
- [Pricing](https://agend.sh/pricing): Current Developer and Pro plans, resource allocations, annual billing, trial limits, and Enterprise inquiries.

## Workflows

- [Interactive prompts and installers](https://agend.sh/use-cases/prompts-sudo): Reproducible installer example with terminal input and output-file verification.
- [REPLs and debugging](https://agend.sh/use-cases/repls-debug): Preserve interpreter and debugger context between agent tool calls.
- [Database operations](https://agend.sh/use-cases/database-operations): Persistent database-client sessions and result verification.
- [SSH and remote systems](https://agend.sh/use-cases/ssh-remote): Access authorized external systems from an agend environment using appropriate network access and credentials.
- [AI agent pipelines](https://agend.sh/use-cases/ai-agent-pipelines): Install dependencies, build, run checks, and debug inside a managed Linux workspace.

## Optional

- [README verification report](https://raw.githubusercontent.com/agend-sh/cli/main/docs/readme-verification.md): Recorded CLI/MCP checks, platform coverage, and known limitations.
- [About agend](https://agend.sh/about): Product focus and company information.
- [Terms of service](https://agend.sh/terms): Subscription, environment, and acceptable-use terms.
- [Privacy policy](https://agend.sh/privacy): Account, billing, environment, and usage-data handling.
- [Data processing addendum](https://agend.sh/dpa): Processing roles, subprocessors, and customer-data provisions.