AgentLink
A coding agent harness built into VS Code
AgentLink runs frontier coding agents with the editor's intelligence, visible execution, reviewable changes, and as much or as little supervision as the task needs.
Why a harness matters
A capable model is only part of a capable coding agent. It also needs a good working environment: editor intelligence instead of text guesses, fast feedback instead of a surprise broken build, and a way for you to supervise or redirect work without taking the controls away.
What makes it different
Better context, better feedback, and better control make agents more useful on real codebases.
The editor is the runtime
Agents get language server navigation, diagnostics, symbols, and workspace-aware rename. Edits arrive as native diffs you can accept, reject, or adjust.
Autonomy is a dial, not a switch
Review every action on risky work, or let familiar tasks move faster with focused rules and Approve for Me. Checkpoints make experiments reversible.
Nothing happens in the dark
Tool calls, approvals, questions, and background agents stay visible in the chat and Activity Shelf, with a browser remote for supervision on the go.
A second opinion from another provider
Route review to a different model provider or external ACP agent, so important work gets an independent set of blind spots.
Context for the code, not the harness
Local codebase retrieval, progressive tool disclosure, bounded terminal output, and session condensing keep the active task in focus.
Bring your own models and workflow
Use ChatGPT/Codex, OpenAI, or compatible providers on your own accounts, with MCP, skills, hooks, and AGENTS.md support out of the box.
One engine, several surfaces
The VS Code extension is the heart of AgentLink. Around it, I'm building supporting surfaces on the same portable core, so you can work with AgentLink in the way other harnesses offer, from a terminal or a desktop app, without losing what makes it different.
VS Code extension
The full experience: language intelligence, native diff review, visible terminals, checkpoints, Fleet, and every approval control.
Browser remote
Check on sessions, answer questions, and inspect read-only diffs from any device, without a remote shell.
Desktop app
A standalone Ask Agent chat that doesn't need VS Code open, and that can switch into your running VS Code sessions.
CLI
A terminal coding agent with full diff review, reviewed commands, MCP, background agents, and optional TypeScript language intelligence.
What stays true on every surface
- Every write is shown as a complete diff before it lands, unless you've chosen to grant more autonomy.
- Commands, approvals, questions, and background work stay visible on every surface.
- Accounts, model preferences, and modes are shared, so switching surfaces doesn't mean starting over.
- Supporting surfaces reuse the same engine rather than reimplementing the agent.
Get started
curl -fsSL https://raw.githubusercontent.com/reefbarman/agentlink/main/scripts/install.sh | bash- 1
Install the extension with the script above, or download a .vsix from the latest release.
- 2
Reload VS Code, open AgentLink from the Activity Bar, and choose Agent.
- 3
Sign in with ChatGPT/Codex, add an OpenAI API key, or configure another compatible provider.
- 4
Start with a bounded request, review the diffs, and use /checkpoint before risky work.
The desktop app is available as an unsigned macOS preview from the desktop-v* GitHub releases. The CLI is in private preview for macOS on Apple Silicon and can be built from source.
What AgentLink is not
It does not train or sell a proprietary model, and there is no cloud middleman between you and your provider.
It is not a VS Code fork, so your extensions, marketplace, and language tooling stay yours.
The desktop app and CLI support the extension rather than replace it. They're still previews, and the extension remains the most complete experience.
The browser remote is for supervision only: diffs are read only and there is no remote shell.
Rolling out coding agents across your team?
I help engineering teams adopt AI coding agents with the right tooling, guardrails, and workflows, whether that is AgentLink or another harness.
- Tooling and harness selection
- Shared AGENTS.md, skills, and MCP setup
- Review practices and guardrails