El-cro
Download
Navigate El-croTap outside or press Esc to close
Product
Roadmap
Organizations
Resources
Pricing
Download El-croSee pricing
El-cro/Product
Available now

The local-first AI editor

One editor from first thought to reviewed change.

El-cro combines a familiar VS Code-based foundation with agentic chat modes, semantic codebase search, inline completions, targeted edits, direct model access, MCP tools, and file checkpoints.

Download El-cro See pricing
01Agentic chat modes
02Semantic indexing
03Direct providers
01

Work at the right level

Use Tab for small continuations, Quick Edit for a selection, or pick a chat mode for the autonomy you want.

  • Tab completions
  • Quick Edit and Apply
  • Ask, Plan, Agent, Debug, and Multitask modes
02

Keep control of every change

Edits stream into the editor as diffs. Review changes individually, accept a group, reject an experiment, or restore a checkpoint.

03

Connect your own stack

Choose hosted or local models, index the codebase for semantic search, and extend Agent with MCP tools. El-cro does not require an El-cro-hosted inference service.

Explore Product

Go deeper.

Available now

Agent

El-cro Agent can inspect the workspace, search files and the semantic index, edit code, read diagnostics, run temporary commands, and work with persistent terminals under your approval controls.

Open page Available now

Semantic Indexing

El-cro builds an incremental local semantic index so Agent can search by meaning, not only exact text. You choose the embedding provider; indexing stays on your machine unless you opt into a remote embedding endpoint.

Open page Available now

Tab

El-cro Tab uses fill-in-the-middle models with debounce, streaming partial results, caching, cancellation, and acceptance tracking so inline suggestions stay responsive.

Open page Available now

Browser

Agent Browser gives El-cro a real Chromium session for local web apps: navigate, snapshot interactive refs, click, type, fill forms, wait, screenshot, and read console or network evidence under Browser tool approvals.

Open page Experimental

Multi-agent

Multitask mode runs read-only subagents in parallel via run_subagent, and Agent can create, list, and remove Git worktrees so write tracks stay isolated until you merge reviewable branches.

Open page Experimental

CLI

The El-cro desktop CLI accepts elcro agent "prompt" and opens the IDE on an Agent chat via an elcro:// deep link, reusing the same providers, tools, and approvals as the sidebar.

Open page Available now

Quick Edit

Select code, invoke Quick Edit, and stream a focused transformation into the editor. The same edit engine powers Apply and Agent edit tools.

Open page Available now

Models

Configure separate models for Chat, Autocomplete, Quick Edit, and Apply across hosted providers, local runtimes, remote self-hosted servers, and OpenAI-compatible endpoints.

Open page Available now

MCP

Configure MCP servers and expose their tools to Agent over local stdio, Streamable HTTP, or SSE transports.

Open page Available now

Checkpoints

Chat checkpoints preserve AI-touched file contents and diff state so you can move backward or forward through an agent session.

Open page Available now

Privacy

An El-cro account unlocks Agent, Tab, and Quick Edit, but model requests never route through an El-cro-managed inference backend, and El-cro sends no product analytics.

Open page
AI coding, under your control

Build with your tools, models, and rules.

El-cro puts you in control of your models, keys, and data.

Download El-croSee pricing