Skip to content

v1.2.0: BYOK (Bring Your Own Key) Configuration for GitHub Copilot 🔑

Choose a tag to compare

View all tags
@GulajavaMinistudio GulajavaMinistudio released this 03 Jul 13:30
v1.2.0
c33453d

Release v1.2.0: BYOK (Bring Your Own Key) Configuration for GitHub Copilot 🔑

Version 1.2.0 introduces a brand-new byok-copilot-config/ module that lets developers plug their own API keys from third-party model providers directly into the GitHub Copilot chat experience in Visual Studio Code. This release is purely additive — no breaking changes to existing agents, skills, or instructions.

🌟 Major Highlights & New Features

🔑 New byok-copilot-config/ Module

A new, self-contained folder is now shipped at the repository root to help developers use BYOK (Bring Your Own Key) with GitHub Copilot in VS Code. The module ships complete documentation, a real-world chatLanguageModels.json template, and screenshots of the actual configuration flow.

  • byok-copilot-config/byok-config-copilot.md — full-length BYOK guide covering built-in providers, extensions, custom endpoints, the chatLanguageModels.json schema, the model property reference, and security best practices.
  • byok-copilot-config/README.md — short entry point with the "How to Use" flow, prerequisites, folder contents table, and security notes.
  • byok-copilot-config/img/ — seven annotated screenshot files covering three VS Code UI screens (Add Models dropdown, Language Models editor, Chat model picker) with multiple states of each (empty dropdown, populated editor, pinned favorites, Other Models section, etc.) used throughout the documentation.

🤝 Multi-Provider Reference Configuration

The shipped chatLanguageModels.json example combines eight provider groups into one ready-to-paste configuration, covering both free and paid models:

  • OpenRouter (vendor: openrouter) — built-in vendor; models are picked up automatically from the OpenRouter account.
  • Copilot & Copilot CLI (vendor: copilot) — built-in Copilot models, with per-model reasoningEffort overrides configured via the settings block.
  • Deepseek Platform — custom endpoint (chat-completions) with Deepseek v4 Flash and Deepseek v4 Pro.
  • Kilo Gateway — custom endpoint exposing Kilo Auto Free.
  • Opencode Zen — custom endpoint with DeepSeek V4 Flash Free, MiMo-V2.5 Free, Big Pickle Free, Nemotron 3 Ultra Free.
  • OpenRouter-AutoFree — custom endpoint pointing at openrouter/free.
  • Opencode Go — custom endpoint with 11 models, including GLM-5.2, Kimi K2.6, Kimi K2.7 Code, DeepSeek V4 Pro, DeepSeek V4 Flash, MiMo-V2.5, MiMo-V2.5-Pro, MiniMax M3, MiniMax M2.7, Qwen3.7 Max, Qwen3.7 Plus.
  • Four of the Opencode Go models (MiniMax M3, MiniMax M2.7, Qwen3.7 Max, Qwen3.7 Plus) override the provider-level apiType to messages so they correctly target Anthropic-style endpoints.

📚 Documentation Coverage

The new BYOK guide is structured as a complete reference — every screenshot is embedded directly next to the step it illustrates, and the model property table covers id, name, url, apiType, toolCalling, vision, maxInputTokens, maxOutputTokens, editTools, thinking, streaming, zeroDataRetentionEnabled, supportsReasoningEffort, reasoningEffortFormat, and requestHeaders.

Other documented areas include:

  • Adding models from built-in providers (Azure, Anthropic, Gemini, OpenAI) and from the additional vendors shown in the Add Models dropdown (OpenRouter, xAI).
  • Adding models via Marketplace extensions (Foundry Toolkit, AI Toolkit, Ollama).
  • Adding models via the Custom Endpoint provider (replacing the deprecated "OpenAI Compatible" option) with all three supported API types: chat-completions, responses, messages.
  • Updating an existing provider's API key or endpoint URL from the Language Models editor.
  • Configuring inlineChat.defaultModel, the inline suggestions model, and the two utility model settings (chat.utilityModel and chat.utilitySmallModel).
  • A dedicated FAQ section covering Copilot Business/Enterprise enablement, local self-hosted models, fully-offline usage, and BYOK without a Copilot plan.

🛡️ Security-First Defaults

A dedicated Security Notes section is now part of the BYOK module. The release ships with these rules baked into both the main guide and the folder README:

  • All API keys in the example file are written as ${input:chat.lm.secret.XXXX} — a VS Code Secret Input pattern that resolves through Secret Storage, never in plaintext.
  • Users are explicitly instructed to redact any real secret IDs before sharing the configuration to a public repository.
  • The guide recommends adding a personal chatLanguageModels.json to .gitignore if it diverges from the template.
  • Copilot Business/Enterprise users are reminded that the organization administrator must enable the Bring Your Own Language Model Key in VS Code policy before BYOK is usable.

🧭 Root README.md Integration

The root README.md has been updated so the new module is discoverable from the very first scroll:

  • A new 🔑 BYOK Copilot Config bullet was added to the Overview list, sitting alongside Custom Agents, Skills, Rules, and Multi-Platform.
  • A new ## 🔑 BYOK Copilot Config section now points readers straight at byok-copilot-config/ for the full setup guide.
  • A > [!NOTE] callout clarifies that BYOK models work without a GitHub account or Copilot plan, and only cover chat + utility tasks (semantic search, inline suggestions, and embeddings still require GitHub Copilot).

📝 Enhancements & Synchronization

  • Heading consistency: Fixed corrupted UTF-8 emoji bytes on two root README.md headings (## 🔑 BYOK Copilot Config and ## 🛠️ Customization) so they render correctly across Markdown viewers.
  • TOC ordering: The byok-config-copilot.md Table of Contents now matches the document body — Security Notes appears immediately before Model Property Reference, matching the actual section order.
  • Folder Contents table: The byok-copilot-config/README.md table now lists the img/ folder explicitly, so newcomers immediately know where the screenshots live.
  • Screenshots added: Seven annotated screenshot files (screen1.png, screen1a.png, screen2.png, screen2a.png, screen3.png, screen4.png, screen5.png) are now embedded directly in the documentation at the exact steps they illustrate.
  • No agent or skill behavior changed: This release only adds new documentation and a configuration template — existing agents, skills, instructions, and prompts are untouched.

📦 Installation & Upgrade Guide

To upgrade your existing project or set up v1.2.0:

  1. Pull the latest changes:

    git pull origin main

  2. (Optional) Copy the new byok-copilot-config/ folder into your project if you want to use it as a local reference, or browse it directly on GitHub.

  3. Open byok-copilot-config/README.md first, then follow the step-by-step guide in byok-copilot-config/byok-config-copilot.md.

  4. Copy the example chatLanguageModels.json from the Real-World Configuration Example section into your VS Code chatLanguageModels.json file:

    • Windows: %APPDATA%\Code\User\chatLanguageModels.json
    • macOS: ~/Library/Application Support/Code/User/chatLanguageModels.json
    • Linux: ~/.config/Code/User/chatLanguageModels.json

  5. Replace each ${input:chat.lm.secret.XXXX} placeholder with your own VS Code Secret Input ID. VS Code will prompt you for the API key the first time each provider is loaded.

  6. Restart VS Code if newly added models do not appear immediately in the model picker.

No code, agent, or skill files were changed in v1.2.0, so existing setups remain fully compatible.

Full Changelog: v1.1.0...v1.2.0

Assets 2
Loading