ouyangbo/EvoScientist-WebUI
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

feat: update documentation and improve comments for clarity across multiple files

Browse Source
This commit is contained in:
X-iZhang
2026-06-27 22:05:42 +01:00
parent 4533b67f18
commit 40bd964051
9 changed files with 34 additions and 26 deletions
Download Patch File Download Diff File Expand all files Collapse all files
CONTRIBUTING.md
+15 -13
View File
@@ -73,18 +73,20 @@ EvoScientist-WebUI/
scripts/ # assemble-standalone.mjs (packs dist/, strips *.map)
src/
app/
api/ # thin same-origin server routes
evosci-config/ # detect backend port
skills/ # install / list / remove skills
memory/ # read / edit global memory
workspace/ # browse / edit / download files
components/ # React components (chat, panels, dialogs)
hooks/ # useChat, useThreads, useAsyncAgents, …
page.tsx # multi-view shell (routes via nuqs)
components/ # shared UI primitives
lib/ # client + server helpers (asyncAgents, memory, …)
server/ # server-only fs helpers with path guards
providers/ # Theme / Client providers
api/ # thin same-origin server routes (browser → local disk)
evosci-config/ # detect backend port from EvoScientist config
skills/ # list / install / uninstall + remote EvoSkills catalog
memory/ # global memory CRUD + observation graph + executions
workspace/ # browse / read / edit / upload / download-zip files
components/ # feature UI: chat, inspector, memory, schedule, skills, dialogs
hooks/ # useChat, useThreads, useAsyncAgents, useScheduledTasks,
# useAvailableModels, useMemoryActivity, useAutoNotify
page.tsx # three-column shell (thread · chat · inspector); nuqs routing
components/ui/ # shared shadcn/ui primitives
lib/ # client helpers (asyncAgents, cronUtils, modelCommand,
# observationGraph, fileLink, summarization, …)
server/ # server-only fs helpers (workspace / memory / skills) w/ path guards
providers/ # ClientProvider (SDK) · ChatProvider · ThemeProvider
```
## Architecture & connection contract
@@ -95,7 +97,7 @@ Three independent layers:
2. **Next `/api/` routes** (server tool layer, same-origin) — the reason distribution is `output: "standalone"` rather than a static export.
3. **Backend** (EvoScientist's `langgraph dev`, default port `6174`) — the anchor process: runs async agents, holds state, serves the SDK.
The backend exposes graphs `EvoScientist` (main, UI-locked), `writing-agent`, `data-analysis-agent`, and the `evomemory` workers. The UI filters thread lists by `metadata.graph_id == "EvoScientist"` so worker/sub-agent threads stay hidden.
The backend exposes graphs `EvoScientist` (main, UI-locked), `writing-agent`, `data-analysis-agent`, the `scheduler` graph (LangGraph crons behind **Scheduled Tasks**), and the `evomemory` workers. The UI filters thread lists by `metadata.graph_id == "EvoScientist"` so worker/sub-agent threads stay hidden.
## Code style
README.md
+8 -2
View File
@@ -28,12 +28,17 @@
## ✨ Features
- **💬 Streaming Chat** — Real-time responses with Markdown, GFM tables, code highlighting, and collapsible thinking/reasoning blocks.
- **💬 Streaming Chat** — Real-time responses with Markdown, GFM tables, math (KaTeX), code highlighting, zoomable Mermaid diagrams, and collapsible thinking/reasoning blocks.
- **🧬 Per-Thread Model Picker** — Switch models per conversation with the `/model` command or a clickable model pill; the choice is persisted to the thread and folded into the next run.
- **👋 Human-in-the-Loop** — Approve / reject / edit tool calls, and answer the agent's structured questions (text + multiple-choice) inline.
- **⚡ Per-Thread Auto-Approve** — Persisted per conversation; survives view and thread switches and reloads.
- **⌨️ Message Queue** — Type while the agent is busy: queue, edit, reorder, steer, or drain follow-up messages without interrupting the active run.
- **🤖 Sub-Agent Activity** — Live step tracking for sub-agents, rendered exactly like the main agent (tool calls + paired results + tables).
- **🗂️ Workspace Browser** — Tree and by-type (Papers / Figures / Data / Code) views with preview, edit, download, and zip-all.
- **🧠 Memory Browser** — View, edit, and manage EvoScientist's global cross-session memory, with "recently updated" highlights and a nav badge.
- **🔗 Click-to-Open File Links** — File paths in agent output are clickable — open them straight from chat in a workspace or memory viewer.
- **🧠 EvoMemory Browser** — EvoScientist's global cross-session memory across three tabs: **Identity** (editable profile files), **Knowledge** (an interactive force-directed observation graph), and **History** (execution + observation timeline).
- **⏰ Scheduled Tasks** — Schedule recurring research runs (daily / weekly / monthly / custom cron) with a visual builder, templates, and Run-now — backed by LangGraph crons.
- **📊 Research Dashboard** — The chat's empty state surfaces recent memory activity, scheduled tasks, threads, and files, with one-click jump-in.
- **🔌 Skills Marketplace** — Install, update, and uninstall the official [EvoSkills](https://github.com/EvoScientist/EvoSkills) catalog with version detection and a detail dialog.
- **📡 Agents Monitor Board** — Watch async background agents (writing / data-analysis) with real run status, live duration, and a side-chat for direct worker debugging.
- **🔁 Async Agent Communication** — Optional per-thread auto-report loops finished background results back to the main agent.
@@ -43,6 +48,7 @@
## 📖 Table of Contents
- [✨ Features](#-features)
- [📦 Prerequisites](#-prerequisites)
- [⚡ Quick Start](#-quick-start)
- [🔑 Configuration](#-configuration)
scripts/assemble-standalone.mjs
+2 -2
View File
@@ -27,8 +27,8 @@ if (existsSync(PUBLIC)) {
}
// Next copies the whole project root into the standalone bundle. Prune it down
// to just the runtime essentials (drops src/, configs, and local notes like
// CLAUDE.md/AGENTS.md so they never get published).
// to just the runtime essentials (drops src/, configs, and local notes so
// they never get published).
const KEEP = new Set([
"server.js",
".next",
src/app/components/AgentsPanel.tsx
+1 -1
View File
@@ -122,7 +122,7 @@ export function AgentsPanel({ onReportToMainChat }: AgentsPanelProps) {
// Hidden power-user feature: a tiny composer per expanded task that sends a
// message straight to that sub-agent's own thread. Note this is a SIDE channel
// — the reply lands in the sub-agent's thread (shown here), the main agent
// doesn't see it (see CLAUDE.md §3 Agents board: no auto loop-back to parent).
// doesn't see it — there is no auto loop-back to the parent agent.
const [chatInput, setChatInput] = useState<Record<string, string>>({});
const [chatBusy, setChatBusy] = useState<Record<string, boolean>>({});
const [chatError, setChatError] = useState<Record<string, string | null>>({});
src/app/components/ChatInterface.tsx
+3 -3
View File
@@ -289,7 +289,7 @@ export const ChatInterface = React.memo<ChatInterfaceProps>(
cancelled = true;
};
}, []);
// Messages typed while the agent is busy (Claude Code-style queue). They
// Messages typed while the agent is busy are queued. They
// drain one-per-idle-window into the thread once it's free. A ref mirrors the
// latest queue so event handlers (key ↑, edit) read current state without
// being recreated; queueIdRef hands out stable keys.
@@ -857,7 +857,7 @@ export const ChatInterface = React.memo<ChatInterfaceProps>(
// Can't compose with no assistant, a pending interrupt, or files still
// uploading. (Unlike before, isLoading is NOT a blocker — see below.)
if (!assistant || hasPendingInterrupt || isUploadingFiles) return;
// Agent busy → queue it (Claude Code style). The queue drains and sends
// Agent busy → queue it. The queue drains and sends
// automatically once this turn finishes (or is stopped); the message is
// appended as the next turn — it never replaces what's already running.
if (isLoading) {
@@ -912,7 +912,7 @@ export const ChatInterface = React.memo<ChatInterfaceProps>(
});
}, []);
// Codex-style "Steer": prioritize this instruction without interrupting the
// "Steer": prioritize this instruction without interrupting the
// active run. It remains in our visible queue and drains in the next idle
// window, which preserves the same non-interrupting contract.
const steerQueuedMessage = useCallback((id: number) => {
src/app/components/HealthIndicator.tsx
+1 -1
View File
@@ -21,7 +21,7 @@ const POLL_INTERVAL_MS = 10_000;
const REQUEST_TIMEOUT_MS = 4_000;
// Theme tokens (light + dark defined in globals.css). The base's shadcn
// primary/secondary background tokens are dead in this fork (see CLAUDE.md), so
// primary/secondary background tokens are dead in this fork, so
// we reference CSS vars directly via arbitrary-value classes (the entries below).
// NOTE: never put a bracketed class literal in a comment — Tailwind's content
// scanner picks it up and emits real (sometimes invalid) CSS.
src/app/globals.css
+2 -2
View File
@@ -10,8 +10,8 @@
:root {
/* App-specific color variables */
/* App-specific colors — warm "paper" palette in the Claude Code desktop
style. Light mode: ivory background, true-white cards, warm charcoal text. */
/* App-specific colors — warm "paper" palette. Light mode: ivory
background, true-white cards, warm charcoal text. */
--color-primary: #33302a;
--color-user-message: #0e5c6e;
--color-user-message-bg: #e6f2f5;
src/lib/asyncAgents.ts
+1 -1
View File
@@ -172,7 +172,7 @@ export function formatAsyncUpdateMessage(task: AsyncTaskReportTarget): string {
}
// Theme dot + label + pulse per status. CSS vars referenced via arbitrary-value
// classes (the base's semantic bg tokens are dead in this fork — see CLAUDE.md).
// classes (the base's semantic bg tokens are dead in this fork).
// NOTE: never put a bracketed class literal in a comment (Tailwind scans those).
export const ASYNC_STATUS_META: Record<
AsyncAgentStatus,
tailwind.config.mjs
+1 -1
View File
@@ -5,7 +5,7 @@ import tailwindcssAnimate from "tailwindcss-animate";
/** @type {import('tailwindcss').Config} */
export default {
content: ["./index.html", "./src/**/*.{js,ts,jsx,tsx}"],
content: ["./src/**/*.{js,ts,jsx,tsx}"],
darkMode: "class",
theme: {
extend: {