A rock-solid agent foundation · The TUI client for DSH · A single Rust file
Why CLAT
CLAT brings the model into your terminal: reading code, editing files, and running commands all happen locally — edits land atomically (batch changes to one file validate and commit in a single snapshot, all-or-nothing), workspace changes export as a hash-manifested Git delivery patch, and every step stays visible, controllable, and auditable.
Sessions persist as append-only logs; settings, credentials, and trust decisions live in local JSON files under ~/.clat. No cloud dependency — your data never leaves your machine.
A single Rust executable with no runtime and nothing to install — it runs out of the box on six platforms across macOS, Linux, and Windows, no Node or Python required.
A trust gate guards every new directory; permissions switch between read-only, project-write, and full-access — every write and command asks first, and file changes stay traceable and recoverable.
Positioning
Spring, DeepSeek Harness, and CLAT share the same "composable container" organ — systems assembled declaratively from swappable parts, an idea reaching back to IoC and OSGi — yet they serve different users and head in different directions. This table orients first-time visitors.
| Spring | DSH | CLAT | |
|---|---|---|---|
| Positioning | Enterprise Java application framework | An "everything is a plugin" agent runtime | A rock-solid agent foundation |
| Users | Programmers building applications | Developers driving agents, and the models themselves | Users who want local-first, single-binary, auditable tooling |
| Container philosophy | Let programmers forget the container | Let the runtime be rewritten at any time | Let the user audit the container |
| Composition time | Compile time / deploy time | Run time | Compile time (static manifest); extensions move out via MCP |
| Extension | jar + interfaces | In-process TypeScript plugins | MCP protocol + gated WASM |
Spring lets programmers forget the container, DeepSeek Harness lets the model rewrite it, and CLAT lets the user audit it. If what you want is a stable, small, see-through agent foundation — this is it.
Features
01 — Models
DeepSeek, GLM, Qwen, Kimi, and Tencent Hy presets are built in, and any OpenAI-compatible endpoint works too. /model opens a two-level picker — vendor first, then model — with keys in per-vendor slots, thinking levels, and live quota telemetry.
02 — Multimodal
Press Ctrl+V to paste a screenshot, drag a path in, or /paste-image from the clipboard — images ride along as structured attachments in both the TUI and the workbench. view_image lets the model inspect project images itself; a model that can't see fails loudly, never silently.
03 — Interaction
The model stops to ask when a decision is yours; steer an active run any time, or press Ctrl+Q to queue a message for the next turn — dispatched automatically. Reasoning renders inline as a collapsible Think row (Ctrl+R), and next-step suggestions appear automatically after each turn — sent only when adopted.
04 — Runtime
No turn limit — a run continues until the task is done; in Plan Mode the model plans read-only first and executes only after your approval. At 80% of the context window the conversation compacts automatically with the original history preserved; Esc cancels any time.
05 — Permissions
Read-only, project-write (default), and full-access, switchable any time via /perm or right from a permission prompt, remembered per project. In the lower two modes file writes and commands still ask every time; commands running sandboxed are approved automatically.
06 — Integration
MCP servers over stdio or Streamable HTTP; GLM Coding Plan mounts the four official GLM servers automatically, /mcp inspects every connection and tool. /web configures the native web search and fetch tools, and the official DSH web quartet installs from the market.
07 — Agent
Memory is explicit and yours: /memory stores project or global knowledge, the model only reads it — conversations never become memory on their own. /jobs and /schedule run background work with time- or completion-based triggers, progress inspectable any time.
08 — Persistence
Every conversation is written to an append-only log, flushed before the model is called; after a crash it recovers to the last complete batch. Logs use the DSH-compatible format readable by other tools, /history full-text-searches committed sessions, and input history is shared across clients.
09 — Skills
Skills layer across the project's .clat/skills/, the user directory, and built-ins; the model loads instructions by name (/skill invokes directly) — read-only, never executing code. Configure ~/.clat/lsp.json for read-only LSP: definition, references, implementation, hover.
10 — Plugins
The signed plugin marketplace pi.at.cn is live: clat plugin and the /plugin UI (identical in TUI and workbench) browse, install, configure, and roll back, with signature checks built in. Manifest v2 lets WASM plugins go online; the DSH ecosystem interoperates.
11 — Ecosystem
One clat dsh connects to a local DSH host and the CLAT interface becomes its terminal: DSH sessions, questions, and approvals all happen in the same TUI with automatic reconnection. It never writes to ~/.dsh — everything goes through the official API, and host feedback renders natively.
12 — Frontends
clat starts a background host with the TUI, web workbench, and WeChat remote on the same sessions; the desktop app installs standalone and shares Host data. The workbench ships a terminal, a code editor, and full Git source control, in three languages.
Get started
The one-line scripts are recommended: they detect your platform, download the release binary, verify the Minisign signature and SHA-256, and put it into ~/.local/bin (Windows: %LOCALAPPDATA%\Programs\clat). First-install verification needs minisign (macOS: brew install minisign); when no prebuilt binary exists the scripts fall back to building from source.
You can also grab the binary for your platform from the Releases page and drop it into your PATH. Installed users self-update with clat upgrade.
| Platform | Asset | Architecture |
|---|---|---|
| macOS | clat-v*-aarch64-apple-darwin.tar.gz | Apple Silicon |
| macOS | clat-v*-x86_64-apple-darwin.tar.gz | Intel |
| Linux | clat-v*-x86_64-unknown-linux-gnu.tar.gz | x64 |
| Linux | clat-v*-aarch64-unknown-linux-gnu.tar.gz | ARM64 |
| Windows | clat-v*-x86_64-pc-windows-msvc.zip | x64 |
| Windows | clat-v*-aarch64-pc-windows-msvc.zip | ARM64 |
Linux builds require glibc 2.39+ (Ubuntu 24.04 generation or newer); older distributions should build from source. After extracting, try clat --help, then run clat inside a project directory — the first run asks you to trust it. The scripts are the same: once installed, just run clat.