Reasonix
A DeepSeek-native coding agent for your terminal, browser, and ACP-compatible editor.
What is Reasonix?
Reasonix is an open-source AI coding agent designed for the terminal, browser, and ACP-compatible editors, engineered around DeepSeek. It uses an append-only loop aligned with DeepSeek's byte-stable prefix cache, achieving 90%+ cache hit on long sessions and reducing input-token costs to about 1/5. The same engine powers the CLI/TUI, desktop app, local browser UI, and ACP-compatible editors. Reasonix is MIT-licensed and developed in public with 97 contributors.
What can Reasonix do?
- 01
Cache-first loop
Append-only history aligned to DeepSeek's prefix cache; 90%+ hit on long sessions; input tokens bill at ~1/5.
- 02
Single Go binary
CGO-free, cross-compiled for darwin/linux/windows × amd64/arm64; no Node runtime needed.
- 03
MCP first-class
Supports stdio, SSE, streamable HTTP; external servers merge tools into one registry under a prefix.
- 04
Plan mode & sandbox
/plan asks model to plan first; permissions and workspace sandbox govern every tool call.
- 05
Subagents & skills
Includes explore/research/review/security-review agents and Markdown skill scripts with isolated tools.
- 06
Terminal, browser, or editor
Full-screen TUI, local browser UI via 'reasonix serve', VS Code extension, or ACP-compatible editors.
Use Cases
- Developers — Run long terminal coding sessions on a DeepSeek-native agent whose append-only, cache-aligned loop holds 90%+ cache hits and cuts input-token cost to about a fifth.
- Terminal users — Install a single CGO-free Go binary across macOS, Linux and Windows with no Node runtime to set up.
- AI coding users — Merge external MCP servers' tools into one registry and use plan mode with a workspace sandbox to govern each tool call.
- Editor-based developers — Work from a full-screen TUI, a local browser UI, the VS Code extension or any ACP-compatible editor.
Quick facts about Reasonix
- Domain Rating
- Platforms
- Web
- Languages
- English
Frequently Asked Questions
Reasonix Traffic Analysis
Alternatives to Reasonix
Looking for a Reasonix alternative? Compare these curated AI tools that offer similar features and use cases.
Prelint
Prelint reviews every pull request against your product specifications, catching product drift, inconsistencies, and misalignment before they ship. It enforces product intent by checking business logic, compliance, tooling, domain language, scope, and strategic alignment. With tenant-isolated infrastructure and no training on your data, Prelint integrates with GitHub and GitLab, using specs in Markdown, YAML, or structured formats. Pricing is $1 per completed review, with $10 free credits and free for open source.
Heard
Heard is a macOS app that listens to your coding agent's events, understands what matters in context, and turns the useful parts into concise, natural speech. It offers multiple listening modes (Co-pilot, Companion, Focus) and supports multi-agent awareness, allowing you to follow parallel sessions with distinct voices. Heard also includes Heard Power for mobile pairing, enabling you to listen, approve, and reply by voice while away from your Mac. The app is open-source (Apache-2.0) and offers free, Pro, and Power plans.
XHawk
XHawk is a multiplayer AI platform for human-agent collaboration. It enables teams to operate with a shared company brain, coordinating specialized agents across product, analytics, and engineering. Agents run 24x7, execute tasks in batch mode, and every action is audited and remembered, building a compounding knowledge layer.