# EvoScientist

> 🔬 Harness Vibe Research with Self-evolving AI Scientists

- **URL**: https://www.freshcrate.ai/projects/EvoScientist
- **Author**: EvoScientist
- **Category**: AI Agents
- **Latest version**: `v0.1.2` (2026-06-01)
- **License**: Apache-2.0
- **Source**: https://github.com/EvoScientist/EvoScientist
- **Homepage**: https://EvoScientist.ai/
- **Language**: Python
- **GitHub**: 2,796 stars, 159 forks
- **Registry**: github
- **Tags**: `ai-agent`, `ai4science`, `multi-agent-system`, `python`, `vibe-research`

## Description

🔬 Harness Vibe Research with Self-evolving AI Scientists

## Recent releases

| Version | Date | Urgency | Changes |
| --- | --- | --- | --- |
| `v0.1.2` | 2026-06-01 | High | # EvoScientist v0.1.2 — Browser WebUI Mode, Standalone Deploy Server, Background Execution & Sonnet 4.6 Default  G'Day! v0.1.2 opens a new browser surface for EvoScientist and hardens the runtime underneath it. The headline is **WebUI mode (beta)** — one terminal now launches a deploy-style LangGraph server *and* the published `@evoscientist/webui` front-end, built on the new **`EvoSci deploy`** standalone server. We also ship **background process management** so long shell jobs survive past t |
| `v0.1.1` | 2026-05-19 | High | # EvoScientist v0.1.1 — DeltaChannel Upgrade, Tier-Aware Skill Mounts, Status & Elapsed Time, QQ Inline Buttons  G'Day! v0.1.1 is a focused follow-up to the v0.1.0 async-sub-agent release. The headline is the **deepagents 0.5.7 → 0.6.2 upgrade**, which moves message storage to LangGraph 1.2's **DeltaChannel** and required a careful rewrite of our resume path so `/resume` no longer shows "0 messages". The upgrade also enables the new **CodeInterpreterMiddleware** with a research-tuned PTC allow |
| `v0.1.0` | 2026-05-08 | High | # EvoScientist v0.1.0 — Async Sub-Agents, Official Docker Image, Personal WeChat & Sessions DB Compaction  G'Day! v0.1.0 is the biggest release since launch. The headline is **async sub-agents** — `writing-agent` and `data-analysis-agent` now run on a managed `langgraph dev` subprocess, so 30-180s long-runners no longer block your chat, and you get **proactive completion notifications** without polling. We also ship an **official multi-arch Docker image**, a **personal-WeChat (iLink) channel b |
| `v0.0.9` | 2026-04-26 | High | # EvoScientist v0.0.9 — Faster Startup, In-Session Model Switching, Unified CLI Commands & DeepSeek V4 Thinking Fix  G'Day! v0.0.9 focuses on **interactive UX polish and developer ergonomics**. Startup latency drops ~7×, the new **`/model` command** lets you switch models without restarting, **`--resume`** brings session resumption to first-class status, and Rich CLI / TUI / Channels now share a unified slash-command pipeline. We also ship a critical **DeepSeek V4 thinking-mode fix**, MCP reli |
| `v0.0.8` | 2026-04-21 | High | # EvoScientist v0.0.8 — Unified Data Directory, Status Bar, Enhanced Ask-User & Auto-Mode  G'Day! We're excited to release EvoScientist v0.0.8. This release **unifies all application data under `~/.evoscientist/`** (separating config from state per XDG spec), adds a **persistent session status bar** with real-time context window tracking, **enhances ask-user prompts** with multiple-choice and free-text validation, renames `--auto-approve` to **`--auto-mode`** for unattended execution, and ship |
| `v0.0.7` | 2026-04-10 | High | # EvoScientist v0.0.7 — Global Skills Directory, Moonshot/Kimi Providers, ccproxy Fixes & Channel Improvements  G'Day! We're excited to release EvoScientist v0.0.7. This release introduces a **three-tier global skills directory** so installed skills are shared across all workspaces, adds **Moonshot and Kimi Coding Plan** as new LLM providers, fixes **ccproxy Codex OAuth compatibility** for the Responses API path, patches **OpenAI-compatible streaming content flattening**, and ships several cha |
| `v0.0.6` | 2026-04-03 | High | # EvoScientist v0.0.6 — Dynamic Context Management, OpenRouter Reasoning, More Effort Mode, GLM-5.1 & Telegram Fix  G'Day! We're excited to release EvoScientist v0.0.6. This release brings **Dynamic Context Management** with two new LangChain middleware (ContextEditing + Adaptive Tool Selection), an **OpenRouter streaming reasoning fix** with expanded model coverage, a **More Effort code generation mode**, **GLM-5.1 model support** for Zhipu providers, and a **Telegram channel startup fix** — |
| `v0.0.5` | 2026-03-27 | Medium | # EvoScientist v0.0.5 — Context Retry Middleware, OpenAI Relay Config, Feishu Event Loop Fix & Compact Conversation  G'Day! We're excited to release EvoScientist v0.0.5. This release brings a **Context Overflow Retry Middleware** that auto-recovers from provider context limit errors, a **`use_responses_api` config option** for OpenAI-compatible relays, a **Feishu SDK event loop isolation fix** for Linux, and a **`/compact` thread_id injection fix** — along with onboard UX improvements and benc |
| `v0.0.4` | 2026-03-24 | Medium | # EvoScientist v0.0.4 — @ file Mentions, Resume History, Feishu WebSocket & LaTeX Setup  G'Day! We're excited to release EvoScientist v0.0.4. This release brings **`@ file` mention support** for injecting file contents directly into messages, **conversation history display on `/resume`**, a **WebSocket subscription mode for Feishu**, and a **LaTeX environment setup step** in the onboard wizard — along with a `uv tool install` compatibility fix and a TUI slash-command completion bugfix.  --- |
| `v0.0.3` | 2026-03-20 | Low | ## EvoScientist v0.0.3 — Voice Input, New Providers, Smarter TUI  G'Day! We're excited to release EvoScientist v0.0.3. This release adds **voice transcription** across all messaging channels, two new first-class LLM providers (**MiniMax** and **DeepSeek**), an **MCP server browser**, a **skill browser** with tag-based filtering, and a major round of TUI/CLI UX polish — multi-line input, timestamps, update notifications, and more — making EvoScientist more natural to talk to, easier to extend, |

## Dependency audit

- **Score**: 87/100
- **Total deps**: 30
- **Resolved**: 20
- **Unresolved**: 10
- **License conflicts**: 0
- **Warnings**: 3
- **Scanned**: 2026-04-27

## Citation

- HTML: https://www.freshcrate.ai/projects/EvoScientist
- Markdown: https://www.freshcrate.ai/projects/EvoScientist.md
- Dependencies JSON: https://www.freshcrate.ai/api/projects/EvoScientist/deps

_Generated by freshcrate.ai. Indexes github releases for AI-agent ecosystem packages._