Skip to main content

lsp

Live diagnostics (errors/warnings) for code, via Language Server Protocol subprocesses or CLI fallbacks - so an agent editing code can catch mistakes before the user does. Built-in probes cover common languages (Python, TypeScript, Go, Rust, …); those are the target languages for diagnostics, not Digitorn's own implementation language.

Exact tools and parameters

catalog.describe_module lsp returns the real tools and every parameter, from the live registry.

Tools, at a glance: diagnostics, notify_change, reload_app. Note reload_app is the self-build tool: it recompiles the session's own app from disk and makes it live for the next test - refusing if the bundle still has compile errors.

Config worth knowing​

yaml
tools:
  modules:
    lsp:
      config:
        settle_seconds: 10
        disable_builtins: false
        # servers:
        #   pyright:
        #     command: pyright-langserver --stdio
        #     extensions: [".py", ".pyi"]

A working app​

Compiles as-is. The brain uses a Digitorn gateway model (the default - no API key of your own needed); see the note below to use your own key or a local model instead.

yaml
app:
  app_id: code_helper
  name: Code Helper
runtime:
  entry_agent: main
agents:
  - id: main
    system_prompt: |
      Help edit code. After changes, check diagnostics for errors and reload
      the app when needed.
    brain:
      provider: openai
      backend: openai_compat
      model: mimo-v2.5-free   # a model the Digitorn gateway serves
      config:
        api_key: placeholder  # ignored in gateway mode
tools:
  modules:
    lsp: {}
    filesystem: {}
  capabilities:
    grant:
      - module: filesystem
        tools: [read, edit, write, glob, grep]
      - module: lsp
        tools: [diagnostics, notify_change, reload_app]