Coding Agents

The desktop app has a Claude Code tab: a local AI workspace where an agent edits files, runs commands, and reports results back to Liza.

Desktop only

Coding agents run only in the desktop app, since they operate on a local working directory. In the browser the Liza assistant is available, but no coding agent.

What an agent can do

• Read and edit files in a working directory
• Run commands (terminal/bash)
• Use the project context
• Report back to Liza (via MCP): check off tasks and post a summary to the chat

Connect a provider

Settings → Personal → AI & Connections (KI & Verbindungen) → Coding-Agent

In the Coding-Agent tab, pick the provider at the top (one sub-tab per provider) and connect it – either via login or with your own API key. You can switch the provider anytime later in the chat tab.

Claude (Anthropic)

• "Connect with Claude" – connects the desktop app to your Claude account (token stored locally), or
• Manual: paste your own key/setup token (claude setup-token or sk-ant-…). Create a key at console.anthropic.com.

OpenAI / Codex

• "Connect with ChatGPT" – browser login (recommended if you have a ChatGPT subscription, no API key needed), or
• API key from the OpenAI platform.

Mistral (Vibe)

• Prerequisite: install the Vibe CLI once (a separate Python command-line tool) – Liza detects it automatically.
• Add a Mistral API key (create one at console.mistral.ai). Uses the Devstral 2 model as a local coding agent.

Perplexity

A pure web research provider (Sonar models) – it does not execute anything locally and does not edit files. Use it to research with sources, then implement the results with Claude or Codex (switch the provider in the tab).

Define custom agents

In the same tab, under "Eigene Agenten" (Custom agents), you can define your own agents – e.g. a "PR reviewer". They appear in the Code tab as a selectable pill.

For each agent you set:

• Name – e.g. "PR-Reviewer".
• Instructions / role – what the agent should do (and in what order), e.g. "1. Read the diff, 2. Flag risks, 3. Report with fix suggestions".
• Steps (optional) – enforce a fixed order via "+ Schritt" (+ Step).
• Access – how much the agent may do:
  • Research / read-only – read and research only
  • Read + Liza – read + report back to Liza (check off / summary)
  • Coding (full access) – edit files and run commands
• Model – Standard, Sonnet, Opus or Haiku.

Use "+ Agent" to add more and Speichern (Save) to store them. In the Code tab you then pick the agent via its pill.

Connection to Liza & working directory

Two settings apply to all agents:

• Liza MCP key (optional): lets the agent check off tasks in Liza and post summaries. See Developer access for the key.
• Default working directory: the directory the agent works in. Applies to new lists; it can be overridden per task in the chat tab.

"Set up Claude Code once — the connection, the Liza key and the default directory then apply across all lists."

Usage

Open the Claude Code tab (globally or inside a list), set the provider and working directory if needed, and type your task. The easiest way is via handing tasks to AI from lists.