Skip to content

Configuration

Jump to bottom
wildfirebill edited this page Aug 25, 2026 · 2 revisions

Configuration

Settings UI

Open Settings by clicking the ⚙️ icon in the activity bar. All configuration is available through the visual editor.

Editor Settings

Setting Default Description
Auto-save On (1000ms delay) Automatically save files after changes
Minimap On Show minimap in editor
Font size 14px Editor font size
Word wrap Off Wrap long lines

LLM Providers

Configure providers in Settings → LLM Providers. Supported presets:

Provider Default URL Notes
Ollama http://localhost:11434/v1 Recommended for local use
LM Studio http://localhost:1234/v1 Enable local server first
llama.cpp http://localhost:8080/v1 Run llama-server
vLLM http://localhost:8000/v1 OpenAI-compatible API

Custom Provider

  1. Click "+ Add provider"
  2. Enter a name and base URL
  3. (Optional) Enter an API key
  4. Click "Test" to verify the connection

Agent Instructions

Create .localai/system.md in your workspace root to customize the agent's behavior.

# System Instructions

You are a helpful coding assistant working in {{workspace}} on branch {{git_branch}}.

- Always use TypeScript strict mode
- Write tests for new logic
- Follow Conventional Commits format
- Never touch files under vendor/

Template Variables

Use these variables in your system prompt — they're substituted at runtime:

Variable Replaced with
{{workspace}} Current workspace path
{{git_branch}} Current git branch name
{{date}} Today's date (YYYY-MM-DD)
{{time}} Current time (HH:MM:SS)
{{weekday}} Day of the week (Monday, Tuesday, ...)

Presets

Available in Settings → Agent Instructions:

  • Code Assistant — general coding help
  • Security Reviewer — security-focused code review
  • DevOps Helper — infrastructure and deployment
  • Refactor Expert — code cleanup and simplification

Agent Memory

Store persistent notes the agent reads for context:

  • Navigate to Settings → Agent Memory
  • Add notes with a key, content, and category (General, Build, Config, Findings, Decisions)
  • Notes are stored in .localai/memory/<key>.md
  • The agent loads these automatically during runs

Workspace Settings

Settings are saved per-workspace in .localai/ directory:

.localai/
├── system.md          → Agent system prompt
├── settings.json      → Workspace-specific settings
└── skills/            → Project-specific skills
    └── <skill-name>/
        └── SKILL.md

Environment Variables

Variable Description
LOCALAI_HOST Server bind address (default: 0.0.0.0)
LOCALAI_PORT Server port (default: 4801)
LOCALAI_WEB_DIST Path to web UI dist (Docker)
LOCALAI_SKILLS_DIR Path to builtin skills (Docker)

Clone this wiki locally