Elyra
 Elyra The coding agent Elyra Conductor Local project conductor Elyra SQL Anywhere Replication-ready SQL engine
Start here
Quickstart
Using Conductor
Features
Projects & sidebar
Terminals & panes
Command history & insights
Workspaces
Tasks
Runbooks
Files & editor
Git
Worktrees & parallel agents
Database browser
Command palette
Keyboard shortcuts
Elyra integration
Elyra agent
Architecture
Architecture & boundaries
Tauri commands
State & persistence
Operations
Releasing & auto-update
Development
Troubleshooting
Elyra
Elyra
Elyra Conductor
Tauri commands

Tauri commands

The webview talks to the Rust core through Tauri's IPC: invoke('<command>', args) for request/response, and listen('<event>') for streamed events. All commands are registered in src-tauri/src/lib.rs.

Command reference

Command Module Purpose
pty_spawn pty.rs Open a PTY for a pane (id, cwd, cols, rows) and start streaming.
pty_write pty.rs Write input bytes to a PTY.
pty_resize pty.rs Resize a PTY to new cols/rows.
pty_kill pty.rs Terminate a PTY session.
pty_title pty.rs Foreground process name of a PTY (for dynamic tab/pane titles); null for an idle shell.
agent_spawn agent.rs Spawn elyra --mode rpc for an agent tab (host only).
agent_send agent.rs Send a JSONL message to the agent's stdin.
agent_kill agent.rs Terminate an agent process.
list_projects projects.rs Scan the root folder; return projects + git branch.
git_status projects.rs Per-project dirty / ahead / behind state.
git_changes projects.rs List working-tree changes for the commit dialog.
git_commit projects.rs Stage, commit, and optionally push.
git_worktree_list / git_worktree_add / git_worktree_remove projects.rs Parallel isolated worktrees per branch. See Worktrees.
detect_gh / gh_pr_list projects.rs GitHub PR + CI status per branch via the gh CLI.
run_step projects.rs Run one runbook step headless (login shell, timeout) for Verify.
history_add / history_query / history_stats / history_clear history.rs Persistent command history & insights. See Command history.
detect_editors projects.rs Find installed external editors (Zed / VS Code / Cursor).
open_in_editor projects.rs Launch a project in a chosen editor.
detect_elyra projects.rs Resolve the elyra binary via the login shell.
detect_terminal projects.rs Detect the external terminal app (iTerm2 / Terminal).
run_in_external_terminal projects.rs Run a file in the external terminal.
home_dir projects.rs Resolve $HOME for the default root.
list_dir fs.rs Directory listing for the file tree (directories first).
read_file fs.rs Read a file (backs the Monaco editor).
write_file fs.rs Write a file (backs ⌘S).
list_tasks fs.rs Scan a project for runnable tasks. See below.
list_runbooks fs.rs List .conductor/notes/*.md for a project (creates the folder if missing). Backs Runbooks.
write_bytes fs.rs Write raw bytes to a path (binary exports like .xlsx).
list_queries / save_queries fs.rs Per-project, private saved queries in .conductor/queries/ (auto-.gitignored).
db_from_env db.rs Build a DB connection config from a project's .env.
db_connect / db_disconnect db.rs Open / close a MySQL or SQLite connection.
db_tables / db_columns db.rs Schema: list tables / a table's columns.
db_query db.rs Run SQL; returns columns + rows (or rows affected). Backs the database browser.

Events

Event Payload Meaning
pty://exit/<id> exit code (number) The PTY's child process exited.
agent://event/<id> / agent://exit/<id> JSON Streamed agent output / exit.

PTY output is not an event: it streams over a binary Channel<Response> passed to pty_spawn, which ships raw bytes (an ArrayBuffer) straight to xterm.js — far cheaper than JSON-serializing every frame.

list_tasks

invoke('list_tasks', { path: string }): Promise<Task[]>

type Task = {
  label: string;   // e.g. "dev"
  command: string; // e.g. "pnpm run dev"
  source: string;  // "package.json" | "composer.json" | "Makefile" | "justfile"
};

Scans package.json (with package-manager detection by lockfile), composer.json, Makefile/makefile/GNUmakefile, and justfile/Justfile/.justfile. See Tasks for the parsing rules.

Adding a command

  1. Write a #[tauri::command] function in the appropriate module under src-tauri/src/.
  2. Register it in the tauri::generate_handler![ … ] list in lib.rs.
  3. Call it from the frontend with invoke('your_command', { …args }).

Keep new commands within the boundary: transport, filesystem, process launching, and UI support — never model calls or secret handling.

Back to docs