Skip to content

README.md

Latest commit

 

History

History
47 lines (38 loc) · 3.76 KB

README.md

File metadata and controls

47 lines (38 loc) · 3.76 KB
Raw

Selfcoder User Documentation

Welcome to the public documentation for Selfcoder, a local and self-hosted AI coding assistant for VS Code. Selfcoder supports LM Studio, Ollama, vLLM, llama.cpp, and other OpenAI Chat Completions-compatible servers.

Selfcoder is designed for developers who want practical AI assistance inside VS Code while keeping model choice, backend configuration, and workspace context under their own control.

Documentation Map

Guide What it covers
Getting Started Install Selfcoder, connect a supported model server, and send your first message.
Core Workflows Choose between Chat, Plan, and Agent, organize and branch conversations, undo an Agent edit, and use native VS Code integrations.
Plan Mode Investigate a repository and prepare an implementation plan under an enforced read-only boundary.
Slash Commands Use built-in sidepanel commands and custom project commands.
Context and Attachments Choose explicit @ context mentions and understand pins, automatic context, attachments, workspace instructions, and token budgeting.
Models and Backends Configure LM Studio, Ollama, vLLM, llama.cpp, or another OpenAI-compatible endpoint and choose useful models.
Settings and Privacy Configure Selfcoder behavior, privacy mode, history, reasoning, vision, and workspace context.
Troubleshooting Fix common connection, model, native chat, attachment, and response quality issues.

What You Can Do With Selfcoder

  • Chat with models from a dedicated VS Code sidepanel.
  • Use Plan mode to read and search a repository, ask clarifying questions, and design an implementation without changing the workspace.
  • Switch the sidepanel into Agent mode to let a model read files, make edits, and run commands across your workspace, with change tracking and one-click revert.
  • Move from Plan to Agent without losing the session context; implementation begins only after you switch modes and send an Agent request.
  • Use models through LM Studio, Ollama, vLLM, llama.cpp, or another compatible endpoint.
  • Ask questions about the current file, selected code, diagnostics, recent work, or repository changes.
  • Pin files into the conversation context when a task needs specific source files.
  • Select files, folders, symbols, Git changes, terminal output, or codebase search with @ mentions for the next request.
  • Attach text files and images, when supported by the selected model.
  • Run sidepanel slash commands such as /context, /review, /models, /export, /compact, and /init.
  • Add custom Agent mode slash commands with Markdown files in each workspace folder's .opencode/commands/.
  • Use @Selfcoder inside VS Code native chat.
  • Expose eligible local models to VS Code's model picker for native chat and agent-style workflows.
  • Search, rename, and favorite conversations globally or within the current repository, or branch from an earlier assistant response.
  • Revert all edits from the current session or undo only the latest Agent edit.

Recommended First Setup

  1. Install Selfcoder from the latest release.
  2. Install and start LM Studio, Ollama, or an OpenAI-compatible server such as vLLM or llama.cpp.
  3. Make a chat-capable model available through the server.
  4. Open the Selfcoder sidepanel in VS Code.
  5. Select your model.
  6. Ask a small, focused question about the current file.

For the full walkthrough, start with Getting Started.