Tollecode for VS Code
AI coding assistant — chat, plan, and build with multiple AI providers, right inside your editor.
What it does
Tollecode brings the full power of the Tollecode agent into VS Code. Ask questions about your codebase, get the agent to refactor files, generate tests, or debug issues — without leaving your editor.
- Chat with your codebase — the agent reads files, lists directories, and understands your project structure
- PLAN and BUILD modes — PLAN mode reasons read-only; BUILD mode writes files and runs shell commands with your approval
- Multiple AI providers — Claude, GPT-4o, Gemini, Ollama, or any custom OpenAI-compatible endpoint
- Session history — conversations persist across restarts, per workspace
- Live tool visibility — watch the agent work in real time: file reads, writes, shell commands, and sub-agent spawns
- Slash commands —
/mode, /model, /memory, /skills, /usage, /thinking, /clear, and more
- File mentions — type
@path/to/file to reference files with autocomplete
- Workspace memory — the agent remembers project context across sessions
- Skills system — inject reusable behaviours from
.md files in .agent/skills/
- Token usage tracking — per-session and all-time cost breakdowns by provider and model
Installation
- Install from the VS Code Marketplace: search for "Tollecode"
- Or download the
.vsix from Releases and run:
code --install-extension tollecode-<version>.vsix
Quick start
- Open a workspace folder in VS Code
- Click the Tollecode icon in the Activity Bar (⚡)
- Configure a provider via the terminal:
tollecode configure add
- Start chatting — the agent reads your codebase context automatically
Keyboard shortcuts
| Shortcut |
Action |
Ctrl+Shift+T / Cmd+Shift+T |
Open Tollecode panel |
Requirements
- The Tollecode CLI must be installed and on your PATH (
tollecode --version)
- Python 3.10+ for the sidecar process
- macOS, Linux, or Windows
Supported providers
- Anthropic — Claude Opus, Sonnet, Haiku
- OpenAI — GPT-4o, o1, o3, and compatible endpoints
- Ollama — local models, fully offline
- Custom — any OpenAI-compatible API
Commands
All commands are available via the Command Palette (Cmd+Shift+P) under the Tollecode category:
- Open Tollecode — focus the sidebar panel
- New Session — start a fresh conversation
- Focus Input — jump to the message input field
Slash commands
Type / in the chat input to see available commands:
| Command |
Description |
/mode plan\|build |
Switch between PLAN and BUILD modes |
/model |
Open provider and model selector |
/new |
Start a fresh session |
/sessions |
Browse session history |
/memory [query] |
View memory or ask a question |
/skills |
Toggle skills on/off |
/todo |
View current session todos |
/usage |
View token usage and costs |
/thinking 0\|1k\|4k\|10k\|32k |
Set thinking budget |
/clear |
Clear chat history from view |
File mentions
Reference files in your messages by typing @ followed by the file path. The agent will autocomplete paths as you type.
Refactor @src/auth.py to use async/await
License
MIT © Tollesoft
| |