Skip to content

docs: refresh bilingual readmes and agent benchmark figures #480

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Merged
0xKT merged 10 commits into from
Sep 18, 2026
Merged

docs: refresh bilingual readmes and agent benchmark figures #480

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
10 commits
Select commit Hold shift + click to select a range
8d81d37
docs: replace third-party agent tables with an attached image
LivXue Sep 17, 2026
0a81ba7
docs: center the third-party agent image in both readmes
LivXue Sep 17, 2026
8e5f993
docs: size the third-party agent image to 70 percent
LivXue Sep 17, 2026
5196397
docs: refresh the third-party agent overview with cards
LivXue Sep 17, 2026
7d3af9e
docs: refresh and align bilingual readmes
LivXue Sep 17, 2026
0e096c5
docs: add the multi-agent orchestration benchmark chart
LivXue Sep 17, 2026
08e8e9a
docs: expand research and orchestration charts to full width
LivXue Sep 17, 2026
9242447
docs: improve readability of full-width benchmark charts
LivXue Sep 17, 2026
a0cedc4
test(cli): anchor the readme quickstart check on heading text
LivXue Sep 17, 2026
27ea463
docs: rotate the dataagentbench and presentbench charts to columns
LivXue Sep 17, 2026
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
156 changes: 99 additions & 57 deletions README.md
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -25,7 +25,79 @@ Powered by the [EverOS](https://github.com/EverMind-AI/EverOS) memory system, Ra

> Raven is pre-alpha. Interfaces and configuration may change quickly.

## 📊 Benchmarks
<p align="center">
<a href="https://github.com/user-attachments/assets/e333694a-0f4c-4f27-8bfe-8120ff5339a0"><img src="https://github.com/user-attachments/assets/e333694a-0f4c-4f27-8bfe-8120ff5339a0" alt="Multi-Agent Orchestration Benchmark: Node F1, Edge F1, Partial Order Accuracy, and Exact Match Rate" width="100%"></a>
</p>

<p align="center"><em>Raven on the Multi-Agent Orchestration Benchmark</em></p>

## ❯❯ Raven Agents

Raven's modular architecture powers four specialized agents, each assembled from reusable harness components with tools, skills, and workflows tailored to its domain. It can delegate a focused task to one agent or coordinate several agents in a shared workflow.

For example, Raven-Research can gather evidence, Raven-Code can implement an experiment, Raven-Oncall can run and monitor it, and Raven-Design can turn the results into charts and a presentation.

Enable the agents you need during onboarding. See [`agents/README.md`](agents/README.md) for configuration details.

### Raven-Research

**Raven-Research** enables **autonomous deep research** for complex questions, literature reviews, and technical analysis. It delivers clear, structured reports with traceable sources, helping users understand unfamiliar domains, compare alternatives, and make informed decisions.

<p align="center">
<a href="https://github.com/user-attachments/assets/d072a514-58fd-41c7-8fcd-40782ebb4c80"><img src="https://github.com/user-attachments/assets/d072a514-58fd-41c7-8fcd-40782ebb4c80" alt="DeepResearch Mixed: Accuracy, Input Tokens, Output Tokens, and Cost" width="100%"></a>
</p>

<p align="center"><em>Raven-Research on DeepResearch Mixed</em></p>

### Raven-Code

**Raven-Code** enables **agentic software development**, turning requirements into working, tested code. It supports feature implementation, debugging, refactoring, data processing, and data analysis, helping users build new capabilities, resolve issues, and improve code quality while following their project's conventions.

<p align="center">
<a href="https://github.com/user-attachments/assets/a516ebc2-f0b4-47d4-b970-7dfd492adc0e"><img src="https://github.com/user-attachments/assets/a516ebc2-f0b4-47d4-b970-7dfd492adc0e" alt="Coding Benchmarks: SWE-bench Pro, SWE-bench Verified, WorkBuddy-Code Reward, and SWE-Refactor" width="100%"></a>
</p>

<p align="center"><em>Raven-Code on Coding Benchmarks</em></p>

<p align="center">
<a href="https://github.com/user-attachments/assets/75f60aa2-3e2f-42fb-aa51-aa111bf078d0"><img src="https://github.com/user-attachments/assets/75f60aa2-3e2f-42fb-aa51-aa111bf078d0" alt="DataAgentBench (2026-08-24 Live): Raven-Code with Opus-5 achieves 0.8762 Pass@1" width="100%"></a>
</p>

<p align="center"><em>Raven-Code tops on DataAgentBench (2026-08-24 Live)</em></p>

### Raven-Design

**Raven-Design** performs **visual design**, turning ideas and content into polished visual deliverables. It creates presentations, brand assets, charts, diagrams, and web interfaces, refining layout, typography, and visual consistency to help users communicate clearly and bring their ideas to life.

<p align="center">
<a href="https://github.com/user-attachments/assets/c42ff722-8aab-40ec-82e3-0655a859a9d9"><img src="https://github.com/user-attachments/assets/c42ff722-8aab-40ec-82e3-0655a859a9d9" alt="PresentBench: Raven-Design, Claude Code, and public leaderboard scores" width="100%"></a>
</p>

<p align="center"><em>Raven-Design on PresentBench</em></p>

<p align="center">
<a href="https://github.com/user-attachments/assets/061b5818-595b-4392-b9b6-53a1effb8347"><img src="https://github.com/user-attachments/assets/061b5818-595b-4392-b9b6-53a1effb8347" alt="Visual Design: Raven-Design, Claude Code, and Hermes on ArtifactsBench Dashboard, ArtifactsBench SVG, and GDPVal" width="100%"></a>
</p>

<p align="center"><em>Raven-Design on ArtifactsBench and GDPVal</em></p>

### Raven-Oncall

**Raven-Oncall** enables **unattended workflow automation** for experimentation, optimization, and continuous monitoring. It autonomously manages workflows from start to completion, sustaining progress over hours or overnight, delivering results, and involving users only when human judgment is needed.

<p align="center">
<a href="https://github.com/user-attachments/assets/8e2b9721-421c-485c-b1dc-1acf901bcb2c"><img src="https://github.com/user-attachments/assets/8e2b9721-421c-485c-b1dc-1acf901bcb2c" alt="AI4AI (Nanochat 50M Pretraining): Bits Per Byte (BPB), Runtime, Tokens, and Cost" width="100%"></a>
</p>

<p align="center"><em>Raven-Oncall vs. Claude Code on AI4AI (Nanochat 50M Pretraining)</em></p>

<p align="center">
<a href="https://github.com/user-attachments/assets/aafe30ca-6693-4d30-956e-73debf3ef2c9"><img src="https://github.com/user-attachments/assets/aafe30ca-6693-4d30-956e-73debf3ef2c9" alt="AI4S Internal Benchmark: Success Rate, Average Total Runtime, Average Total Tokens, and Average Cost" width="100%"></a>
Comment thread
gloryfromca marked this conversation as resolved.
Resolved
</p>

<p align="center"><em>Raven-Oncall vs. Claude Code on AI4S Internal Benchmark</em></p>

## ❯❯ Benchmarks

| Benchmark | Raven Result | Comparison |
| --- | --- | --- |
Expand All @@ -37,50 +109,15 @@ Results describe the published test configurations; model, task set, and evaluat

https://github.com/user-attachments/assets/3c541dae-5852-447f-8ea6-c9877612ad57

## ❯❯ Connect Third-Party Agents

## 🤝 Raven Agents

Raven's modular architecture powers four state-of-the-art agents, each assembled from reusable harness components with tools, skills, and workflows tailored to its domain. It can delegate a focused task to one agent or coordinate several agents in a shared workflow.

| Agent | What it does |
| --- | --- |
| **Raven-Research** | Searches the live web, reads and compares sources, and produces research reports with citations and references. |
| **Raven-Code** | Writes, runs, and debugs code with state-of-the-art performance, covering feature development, bug fixes, refactoring, scripting, and testing. |
| **Raven-Design** | Creates, edits, and reviews visual work: brand assets, diagrams, charts, illustrations, icons, slide decks, and interface designs. |
| **Raven-Oncall** | Runs and monitors experiments and long-running jobs on local or remote machines, evaluates results, adjusts subsequent runs, and reports the outcome. |

For example, Raven-Research can gather evidence, Raven-Code can implement an experiment, Raven-Oncall can run and monitor it, and Raven-Design can turn the results into charts and a presentation.

Enable the agents you need during onboarding. See [`agents/README.md`](agents/README.md) for configuration details.

## 🔌 Preset Third-Party Agents

Raven includes presets for these third-party agents, so you can bring their capabilities into its orchestration workflows.
Raven can connect to and orchestrate these third-party agents, delegating tasks and coordinating their capabilities in shared workflows.

<table width="100%">
<tr>
<td width="20%" align="center"><img src="ui-web/src/assets/agents/claudecode-color.svg" alt="Claude Code" width="64" height="64"><br><strong>Claude Code</strong></td>
<td width="20%" align="center"><img src="ui-web/src/assets/agents/codex-color.svg" alt="Codex" width="64" height="64"><br><strong>Codex</strong></td>
<td width="20%" align="center"><img src="ui-web/src/assets/agents/opencode.svg" alt="OpenCode" width="64" height="64"><br><strong>OpenCode</strong></td>
<td width="20%" align="center"><img src="ui-web/src/assets/agents/hermesagent.svg" alt="Hermes Agent" width="64" height="64"><br><strong>Hermes Agent</strong></td>
<td width="20%" align="center"><img src="ui-web/src/assets/agents/openclaw-color.svg" alt="OpenClaw" width="64" height="64"><br><strong>OpenClaw</strong></td>
</tr>
<tr>
<td align="center"><img src="ui-web/src/assets/agents/miromind.svg" alt="MiroThinker" width="64" height="64"><br><strong>MiroThinker</strong></td>
<td align="center"><img src="ui-web/src/assets/agents/copilot-color.svg" alt="GitHub Copilot" width="64" height="64"><br><strong>GitHub Copilot</strong></td>
<td align="center"><img src="ui-web/src/assets/agents/qwen-color.svg" alt="Qwen Code" width="64" height="64"><br><strong>Qwen Code</strong></td>
<td align="center"><img src="ui-web/src/assets/agents/codebuddy-color.svg" alt="CodeBuddy" width="64" height="64"><br><strong>CodeBuddy</strong></td>
<td align="center"><img src="ui-web/src/assets/agents/qoder-color.svg" alt="Qoder" width="64" height="64"><br><strong>Qoder</strong></td>
</tr>
<tr>
<td align="center"><img src="ui-web/src/assets/agents/grok.svg" alt="Grok Build" width="64" height="64"><br><strong>Grok Build</strong></td>
<td align="center"><img src="ui-web/src/assets/agents/kimi.svg" alt="Kimi Code" width="64" height="64"><br><strong>Kimi Code</strong></td>
<td align="center"><img src="ui-web/src/assets/agents/pi.svg" alt="Pi" width="64" height="64"><br><strong>Pi</strong></td>
<td colspan="2"></td>
</tr>
</table>
<p align="center">
<img src="https://github.com/user-attachments/assets/3370c883-00ac-4471-97ef-f4312df77202" width="70%" alt="Third-party agents: Claude Code, Codex, OpenCode, Hermes Agent, OpenClaw, MiroThinker, GitHub Copilot, Qwen Code, CodeBuddy, Qoder, Grok Build, Kimi Code, and Pi">
</p>

## 🚀 Quick Start
## ❯❯ Quick Start
Comment thread
gloryfromca marked this conversation as resolved.
Resolved

### 📦 Install

Expand Down Expand Up @@ -124,7 +161,7 @@ checkout reads the tree in place. Setup asks about each product and registers
the ones you take up, on the model it is tuned for or on this raven's LLM.
See [`agents/README.md`](agents/README.md).

## 🏠 Self-Hosting
## ❯❯ Self-Hosting

Raven can run directly from a checkout or as a single Docker Compose service. The
Compose deployment serves the built page through nginx, keeps the Raven engine
Expand All @@ -149,7 +186,7 @@ docker compose up
```

Open <http://127.0.0.1:18793>. The Compose container runs the full `gateway`
engine so providers added from **Settings > Models** are available on the next
engine so providers added from **Settings > Model providers** are available on the next
turn without restarting.

For the detailed container layout, sign-in flow, provider setup, and operational
Expand Down Expand Up @@ -203,12 +240,12 @@ uv run raven web
terminal exits. It defaults to `http://127.0.0.1:18792`. Use
`uv run raven web --foreground` when debugging, or `uv run raven web --stop` to
stop the resident engine. The first run can start without a configured model;
add one from **Settings > Model Providers** or run `uv run raven onboard`.
add one from **Settings > Model providers** or run `uv run raven onboard`.

To run only the engine without the browser launcher, use
`uv run raven gateway`.

## 🧩 Core Systems
## ❯❯ Core Systems

| System | What it adds |
| --- | --- |
Expand All @@ -226,7 +263,7 @@ To run only the engine without the browser launcher, use
</div>


## 🌐 Launch WebUI
## ❯❯ Launch WebUI

Raven's WebUI brings conversations, multi-agent collaboration, and workspace management into your browser. Chat with agents, follow task progress, inspect files and outputs, and browse memory and skills in one place.

Expand All @@ -242,7 +279,7 @@ The command opens the WebUI in your browser and keeps Raven running in the backg

> **Screenshot placeholder 3:** Memory and skill management.

## 📋 Command Reference
## ❯❯ Command Reference

| Command | Purpose |
| --- | --- |
Expand Down Expand Up @@ -280,9 +317,7 @@ The command opens the WebUI in your browser and keeps Raven running in the backg

Run `raven --help` or `raven <command> --help` for the complete CLI surface.



## 📚 Documentation
## ❯❯ Documentation

- [Documentation index](docs/README.md)
- [Developer workflow](docs/dev.md)
Expand All @@ -299,7 +334,7 @@ Run `raven --help` or `raven <command> --help` for the complete CLI surface.

</div>

## 🗂️ Repo layout
## ❯❯ Repo layout

The shared Python runtime lives in `raven/`. Agent definitions, plugin distributions, frontends, and development tools live alongside it.

Expand Down Expand Up @@ -367,7 +402,7 @@ The following runtime packages and modules form the canonical commit scopes unde
| `updates` | Release discovery, upgrade planning, installation handoff, and update notices |
| `utils` | Shared utilities, including atomic file writes |

## 🏗️ Architecture
## ❯❯ Architecture

Each runtime entrance assembles Raven through the same **Assembly Root**, `raven/core/runtime.py:build_runtime`. Configuration and plugin contributions determine the components in a runtime generation; the Spine schedules turns and delivers events around the Agent Loop.

Expand Down Expand Up @@ -408,9 +443,9 @@ See the [Context Map](CONTEXT-MAP.md) for subsystem boundaries, the [Runtime Con

</div>

## 🌐 EverMind Ecosystem
## ❯❯ EverMind Ecosystem

EverMind connects memory research, production-ready products, and practical
[EverMind](https://evermind.ai/) connects memory research, production-ready products, and practical
integrations into one open-source ecosystem.

<table>
Expand Down Expand Up @@ -481,10 +516,17 @@ Together, these projects form EverMind's research-to-runtime stack: methods
and benchmarks become reusable memory infrastructure, products, and agent
integrations.

## 🤝 Contributing
<br>
<div align="right">

[![](https://img.shields.io/badge/-Back_to_top-gray?style=flat-square)](#readme-top)

</div>

## ❯❯ Contributing

Issues and pull requests are welcome. Start with the [developer workflow](docs/dev.md), follow [AGENTS.md](AGENTS.md) for repository rules, and use [GitHub Discussions](https://github.com/EverMind-AI/Raven/discussions) for design conversations.

## ⚖️ License
## ❯❯ License

[Apache License 2.0](LICENSE)
Loading
Loading