Changelog

Latest

Creor follows a rolling release cycle with updates every 1-2 weeks. The editor checks for updates automatically and prompts you to install them. You can also check manually from Help > Check for Updates.

For the complete release history and detailed changelogs, visit the GitHub Releases page.

v0.5.0 — Plan Mode & Background Agents

Released April 2026. This release introduces plan mode for structured multi-step tasks, background agents for parallel work, and significant improvements to the diff viewer.

New features

• Plan Mode — toggle with Shift+Cmd+P. The agent creates a structured plan before writing code, with a dedicated Plan panel for reviewing and approving steps.
• Background Agents — launch multiple agents that work in parallel on separate tasks. Each agent runs in its own git worktree, so changes don't conflict.
• Question Dock — when the agent needs clarification, questions appear in a persistent dock at the bottom of the chat instead of inline, reducing interruptions.
• Session archive — archive old sessions to keep the sidebar clean. Archived sessions are searchable and can be restored.
• Skills system — define reusable prompt templates as skills (e.g., /review-pr, /write-tests) that can be triggered from the chat.

Improvements

• Diff viewer now shows changes grouped by file with inline accept/reject buttons.
• Terminal tool output is rendered in a real terminal emulator with full ANSI color support.
• Model picker redesigned with speed/cost/context indicators and recent model history.
• Chat input now supports multi-line editing with Shift+Enter and @-mention file references.
• Settings UI rewritten with 12 modular sections for easier navigation.

Bug fixes

• Fixed: agent could not read files with non-UTF-8 encodings.
• Fixed: inline edit decorations sometimes persisted after accepting changes.
• Fixed: session list did not update when switching workspaces.
• Fixed: permission dialog blocked the UI when the agent ran multiple tools in parallel.

v0.4.0 — 19 LLM Providers & MCP

Released February 2026. A major expansion of provider support, the introduction of MCP (Model Context Protocol) integration, and the new RAG-powered codebase search.

New features

• 19 LLM providers supported — added Groq, Cohere, xAI, Perplexity, Together AI, DeepInfra, Cerebras, Vercel AI, OpenRouter, and GitLab alongside existing Anthropic, OpenAI, Google, AWS Bedrock, Azure, and Mistral.
• MCP integration — connect external tools and data sources via the Model Context Protocol. Configure MCP servers in settings or install them from the MCP marketplace.
• RAG codebase search — Creor now indexes your codebase using embeddings (Voyage AI or Nomic) for semantic code search. The agent uses this to find relevant code faster.
• Creor Gateway — managed API access to all providers through a single Creor account. No need to manage individual API keys.
• Web search and web fetch tools — the agent can search the web and read documentation pages to answer questions or find solutions.

Improvements

• Multi-edit tool — the agent can now apply multiple edits to the same file in a single operation, reducing round trips.
• Batch tool — tools can run in parallel for faster task completion.
• Provider health indicators — the model picker shows live status for each provider.
• BYOK keys are now stored in the OS keychain instead of plain config files.
• Agent now uses ripgrep for code search, matching VS Code's built-in search speed.

Bug fixes

• Fixed: large file edits sometimes timed out on slow connections.
• Fixed: Google Vertex AI authentication failed when using workload identity.
• Fixed: chat panel did not scroll to latest message in long sessions.

v0.3.0 — AI-Native Editor Launch

Released December 2025. The first public release of Creor, establishing the core AI-integrated editor experience.

Features at launch

• VS Code-compatible editor — full support for VS Code extensions, themes, keybindings, and settings. Familiar interface with AI built in natively.
• AI chat panel — converse with an AI agent that can read, search, edit, and run commands in your project. No extensions or plugins required.
• 25+ built-in tools — file read/write/edit, glob, grep, bash, git integration, code search, LSP integration, and more.
• Inline edit — select code and press Cmd+K to get AI-powered inline suggestions with accept/reject UI.
• Permission system — granular control over what the agent can do. Default ask-before-acting policy with configurable overrides.
• Multi-provider support — Anthropic, OpenAI, Google AI, Google Vertex AI, AWS Bedrock, Azure OpenAI, and Mistral at launch.
• Session management — conversations are saved per-project. Switch between sessions, continue old conversations, or start fresh.
• PTY integration — the agent runs terminal commands in a real PTY with full shell support, not a sandboxed subprocess.

Known limitations at launch

• macOS only (Apple Silicon and Intel). Windows and Linux builds followed in v0.3.1.
• No codebase indexing — the agent relied on grep and glob for code discovery.
• Single session at a time — no parallel agents or background tasks.