Opencodex

A standalone, Codex-style coding agent that runs entirely inside VS Code. Opencodex is currently powered by OpenCode's public free models. No OpenCode account, API key, OmniRoute, or local model server is required.

Open a project, pick a free model, and describe what you want. Opencodex reads files, searches your codebase, applies edits, and runs commands in a visible agent loop and then streams a Markdown answer with its reasoning and work steps shown alongside.

More providers coming soon.

Highlights

• Free and anonymous: Uses OpenCode's live free-only model catalog. Only models currently marked free are offered.
• No account or API key: Works out of the box; the first-launch screen is a simple one-time setup.
• Codex-style agent loop: A collapsible work panel shows each tool step, live activity, and streamed reasoning.
• Safe by default: All agent paths are confined to the open workspace, .env and credential files are blocked, and file edits participate in editor undo.
• Flexible approval modes: Confirm every action, auto-approve edits, or run fully autonomous with auto-approved edits and commands.
• Conversation management: Switch, archive, and permanently delete conversations; Git restore points roll back tracked files to any agent response.
• Resilient streaming: Up to five visible connection retries, animated streamed answers, and persistent red error messages instead of fake Done. responses.

Get started

1. Install the extension from the VS Code Marketplace.
2. Open a project folder and click the Opencodex icon in the Activity Bar.
3. Complete the first-launch settings screen. SearXNG is optional; leave it blank to skip web search.
4. Select a free model from the live model picker below the composer.
5. Start a conversation.

Note on models: There is no automatic model selection. Opencodex remembers the model you explicitly choose. If a model leaves OpenCode's free catalog, the picker resets and asks you to choose another.

Features

Agent loop & interface

• Switchable, archivable project conversations with permanent deletion for archived conversations
• Right-aligned user bubbles and Markdown-formatted assistant responses, with headings, lists, emphasis, links, inline code, and fenced code blocks
• Collapsible Codex-style work panel with an animated loading sheen while the agent reasons or uses tools
• Animated streamed answers with persistent work details for the current run
• Separate work lines for each agent-loop step, plus Markdown-formatted reasoning and work details
• Automatic follow-to-bottom scrolling inside active thought and work panels, during streamed responses, and during tool activity
• Scroll-aware jump-to-latest control with a live activity state
• Message timestamps, copy actions, and Git-tree restore points for Git-tracked projects
• One-item prompt scheduling with remove and Steer controls
• Invisible continuation nudges when a model clearly stops before performing its next stated action

Reliability

• Provider generation failures render as persistent red error messages instead of false Done. responses
• Five provider connection attempts with visible reconnect activity
• Live free-only OpenCode model catalog, refreshed automatically

Settings

First-launch setup covers agent steps, approval mode, and optional SearXNG:

Setting	Default	Description
opencodex.model	(none)	Selected free OpenCode model ID. Choose it from the sidebar.
opencodex.maxSteps	20	Maximum tool-loop steps per request (1-50).
opencodex.approvalMode	ask	ask: confirm every edit and command; edits: auto-approve edits; autonomous: auto-approve edits and commands.
opencodex.searxngUrl	(empty)	Optional base URL for a SearXNG instance with JSON output enabled.

Safety model

• All paths are constrained to the open workspace; files outside it cannot be read, written, or deleted.
• .env and common credential files cannot be read or edited.
• Approval modes can require every confirmation, auto-approve edits, or auto-approve edits and commands.
• Edits use VS Code workspace edits and participate in editor undo.
• Commands always require approval.
• Only anonymous models whose OpenCode IDs are currently marked free are shown.

Web search (optional)

Set a SearXNG base URL in Settings to enable the agent's web_search tool. The server must allow JSON output (format=json).

• http://127.0.0.1:8080 works when the extension host runs directly on your machine.
• Dev Containers and remote workspaces need an address that can reach the host machine, for example its LAN address, or host.docker.internal where supported.

Requirements

• VS Code 1.96.0 or newer
• An internet connection to https://opencode.ai/zen/v1 for the model catalog and completions
• (Optional) A SearXNG instance with JSON output enabled for web search

Known limitations

• Model availability depends on OpenCode's free catalog; the picker is refreshed live and resets if your chosen model is removed.
• Web search requires a user-provided SearXNG instance; no bundled search backend is included.
• Restore points are created only for Git-tracked projects and only for newer responses.

Commands

Command	Description
Opencodex: Focus Chat	Focus the Opencodex chat view.
Opencodex: Open Settings	Open the settings screen.
Opencodex: New Chat	Start a new conversation.

Repository and support

• Bug reports and feedback: open a GitHub Issue and mention the extension version, your VS Code version, and the selected free model when possible.
• Version history: check the repository Releases tab for per-version changes.
• Contributions: open a pull request. The Development section below shows how to build and verify changes locally.
• Model availability changes: OpenCode's free catalog changes over time; refresh the model picker for the current list.

License

MIT. See LICENSE.