Harness Hub
All open source

DeepSeek-Reasonix

Open-source coding agent for terminal, desktop, browser and editor, with a DeepSeek preset.

esengine/DeepSeek-Reasonix

Maintained by esengine

Reasonix is an open-source AI coding assistant that runs on the user's machine. It reads code, edits files and runs commands and tests inside a chosen project folder. A DeepSeek preset is built in, and any OpenAI-compatible service can be added with one config entry, billed directly by the model provider. Permission levels range from asking every time to auto-running everyday actions, and forbid rules always apply. A plan mode waits for approval before acting, and per-turn rewind restores files its edit tools changed. It ships as a desktop app (Reasonix Studio), a full-screen terminal UI, and browser and editor surfaces.

Pros & cons

Each point cites its sources and says what kind of evidence backs it. A second reviewer tried to refute every claim before it was kept.

Pros

  • DocsBuilt around DeepSeek's byte-exact prefix caching, keeping prompts stable to cut cost.[1][2]
  • DocsFour frontends (desktop Studio, TUI, local web, VS Code/ACP) plus file rewind by conversation turn without touching git.[1]
  • DocsMIT, config-driven and multi-model: DeepSeek is only a preset, and MCP servers plug in tools.[1]
  • MetadataVery active: about 36k stars since April 2026, near-daily Studio releases.[1]

Cons

  • User reportsHN users argue you don't need a DeepSeek-native agent: Claude Code or Codex with a bridge already gets most of the cache hits.[1][2]
  • User reportsThe 2.x Studio line is 'still moving'; users ask for a stable release, and desktop context compaction fails.[1][2][3]
  • DocsRewind covers only edits made by its own tools, not changes from shell commands.[1]
  • MetadataThe issue tracker is mostly Chinese-language, with about 541 open issues.[1]
Best for
Budget-conscious developers who use DeepSeek models heavily and want a harness tuned for its cache.
Watch out for
Choose 1.x for stability; 2.x Studio is still changing fast.

Assessed Oct 29, 2026 · adversarially reviewedSuggest a correction

obra/superpowers

An agentic skills framework and software methodology that works.

Maintained by Jesse Vincent (obra)

4 pros · 4 cons

mattpocock/skills

Small, composable agent skills for real engineering work.

Maintained by Matt Pocock (AI Hero)

4 pros · 4 cons

ECC

affaan-m/ECC

Agent harness performance optimization system: skills, instincts, memory and security for coding agents.

Maintained by Affaan Mustafa

4 pros · 4 cons

Connections

Relations from the Harness Hub knowledge graph, as charted in the State of Harness reports.

Maintained by · 1

  • esengineOrganisation

Sources

  1. GitHub repo and README
  2. GitHub API repo metadata