Skip to content
Back to search

OnePagent

One Page Agent.The browser-native workbench that lives in One Page.

Safety scan

A static text scan, not a guarantee. Select the badge to see what was scanned and every finding.

Checked today
Badge for your README
findsafeskills safety rating

Markdown

[![findsafeskills safety rating](/api/listings/sligter--onepagent/badge.svg)](/listings/sligter--onepagent)

HTML

<a href="/listings/sligter--onepagent"><img src="/api/listings/sligter--onepagent/badge.svg" alt="findsafeskills safety rating" /></a>

Install

i

Install instructions from the README

As written by the repo's author. findsafeskills has not verified these steps.

git clone https://github.com/sligter/OnePagent.git
cd OnePagent

Double-click `onepagent.html`, or serve it from any static server:

python -m http.server 8000
# or
npx serve .

Visit `http://localhost:8000/onepagent.html`, click **Settings** in the top bar to configure Provider / API Key / Models. Takes effect immediately.

> **Service Worker note**: the bundled `sw.js` is only a PWA/offline cache for the app shell and static assets. LLM and Tavily requests are browser direct fetches with the configured BYOK credentials; the Service Worker does not proxy, inject, or hide API keys.

---

What was scanned

The files the safety scan read, as written by the repository's author and shown as plain text. Lines with a finding are marked.

README.md · README · 257 lines

View this file on GitHub (read at the repository's current version: the scan did not record which commit it read)

  1. <div align="center">
  2. <img src="logo.svg" alt="OnePagent" width="128" height="128" />
  3. # OnePagent
  4. ### *One Page, Omnipotent Agent.*
  5. **The browser-native AI agent workbench that lives in a single file.**
  6. [![HTML](https://img.shields.io/badge/single--file-HTML-ff6b35?style=flat-square)](onepagent.html)
  7. [![Zero Build](https://img.shields.io/badge/build-none-00ff88?style=flat-square)](#getting-started)
  8. [![BYOK](https://img.shields.io/badge/keys-BYOK-5b8af5?style=flat-square)](#configuration)
  9. [![License](https://img.shields.io/badge/license-MIT-aa66ff?style=flat-square)](#license)
  10. **English** &nbsp;|&nbsp; [简体中文](README.zh.md)
  11. [Live Demo](https://onepagent.top) &nbsp;&middot;&nbsp; [Deploy Your Own](#deploy) &nbsp;&middot;&nbsp; [Configuration](#configuration)
  12. </div>
  13. ---
  14. Open one HTML file and you get a fully-featured, internet-aware, programmable, extensible AI agent — multi-turn chat, tool calls, Python sandbox, web search, skills, context compaction, long-term memory, file operations, cloud sync — all running on a single page.
  15. > No backend. No `npm install`. No Docker. Just one `.html` that carries an entire universe.
  16. ---
  17. ## Preview
  18. ![OnePagent preview](https://jsd.onmicrosoft.cn/gh/mydracula/image@master/20260421/188f31edc79848ff9ed581bc3b5339ff.png)
  19. ---
  20. ## Highlights
  21. | Capability | Description |
  22. |---|---|
  23. | Single-file deployment | Drop `onepagent.html` on any static host or open it locally |
  24. | Multi-provider LLM | Anthropic / OpenAI / DeepSeek with custom endpoints; BYOK keys are stored locally and sent directly to the configured API endpoint |
  25. | Reasoning levels | Inline selector with OpenAI `none / low / medium / high / xhigh / max` tiers (plus Auto) |
  26. | Long context compaction | Per-model context window with automatic LLM-driven summary compression |
  27. | Long-term memory | Opt-in persistent facts / preferences / events / skills across sessions — auto-extraction, agent tools, manual CRUD, tag & keyword search |
  28. | MCP Servers | Paste `mcpServers` JSON to import (`streamable_http` / `sse`), Bearer auth, optional CORS proxy |
  29. | Plan Mode | Agent investigates with read-only tools, drafts a Markdown plan for approval, then executes |
  30. | Ralph Loop | Fully unattended continue-until-done mode with marker, max/unlimited iterations, Stop, and no-progress guard |
  31. | Sub-agents | Delegate bounded read-only research tasks and monitor runs in the side panel |
  32. | Agent Swarm | Opt-in parallel orchestrator-worker fanout: lead emits multiple `SwarmSpawn` calls in one turn, role-scoped workers (researcher / critic / writer / coder) run concurrently with token budgets |
  33. | Human-in-the-loop | Agent can ask for text, choices, or confirmations when a task needs user input |
  34. | TodoWrite | Agent maintains a visible task list (pending / in-progress / completed) |
  35. | Hooks | User-defined JS handlers on 6 agent lifecycle events |
  36. | Python sandbox | Execute Python in-browser via Pyodide |
  37. | Remote sandbox | Optional isolated runtime for heavier code tasks |
  38. | Web Search | Tavily integration with `basic` / `advanced` depth modes |
  39. | Skills system | Install `.skill` / `.zip` packs, pull from GitHub, create in-page, or let AI manage skills with `SkillManager` |
  40. | Conversation management | Multiple sessions, folders, drag-and-drop, IndexedDB persistence, one-click export |
  41. | Cloud Sync | Incremental sync to S3-compatible buckets with optional AES-256-GCM encryption |
  42. ---
  43. ## Architecture
  44. ```
  45. ┌─────────────────────────────────────────────────────────────┐
  46. │ onepagent.html │
  47. │ ┌───────────┐ ┌──────────┐ ┌────────────┐ ┌──────────┐ │
  48. │ │ UI Shell │ │ Chat / │ │ Tools / │ │ Skills │ │
  49. │ │ (3-column)│ │ Streams │ │ MCP Bus │ │ Registry │ │
  50. │ └─────┬─────┘ └────┬─────┘ └─────┬──────┘ └────┬─────┘ │
  51. │ │ │ │ │ │
  52. │ ┌─────┴─────────────┴──────────────┴──────────────┴─────┐ │
  53. │ │ Pretext Layout Engine (inlined) │ │
  54. │ │ markdown → blocks → lines → flowed DOM │ │
  55. │ └──────────────────────────────────────────────────────┘ │
  56. │ │ │
  57. │ ┌─────┴───────┐ ┌──────────────┐ ┌───────────┐ ┌────────┐ │
  58. │ │ Service │ │ LocalStorage │ │ Pyodide │ │ S3 │ │
  59. │ │ Worker │ │ + IndexedDB │ │ (Python) │ │ SigV4 │ │
  60. │ │ (PWA cache) │ │ (all state) │ │ │ │ Client │ │
  61. │ └─────┬───────┘ └──────────────┘ └───────────┘ └───┬────┘ │
  62. └────────┼──────────────────────────────────────────────┼─────┘
  63. │ │
  64. ▼ ▼
  65. ┌──────────────┐ ┌────────────┐ ┌──────────┐ ┌─────────────┐
  66. │ Anthropic │ │ OpenAI │ │ Tavily │ │ Your bucket │
  67. │ DeepSeek │ │ … │ │ │ │ AWS/R2/MinIO│
  68. └──────────────┘ └────────────┘ └──────────┘ └─────────────┘
  69. ```
  70. ---
  71. ## Getting Started
  72. ```bash
  73. git clone https://github.com/sligter/OnePagent.git
  74. cd OnePagent
  75. ```
  76. Double-click `onepagent.html`, or serve it from any static server:
  77. ```bash
  78. python -m http.server 8000
  79. # or
  80. npx serve .
  81. ```
  82. Visit `http://localhost:8000/onepagent.html`, click **Settings** in the top bar to configure Provider / API Key / Models. Takes effect immediately.
  83. > **Service Worker note**: the bundled `sw.js` is only a PWA/offline cache for the app shell and static assets. LLM and Tavily requests are browser direct fetches with the configured BYOK credentials; the Service Worker does not proxy, inject, or hide API keys.
  84. ---
  85. ## Skills
  86. Install skills from the left **Skills** panel via Market, `.skill/.zip`, GitHub, or Create. GitHub folder URLs are supported, e.g. `https://github.com/anthropics/skills/tree/main/skills/skill-creator`.
  87. The agent can also use `SkillManager` to list, install, update, enable/disable, or remove skills; remote installs and destructive/high-risk changes require confirmation.
  88. ---
  89. ## Remote Sandbox
  90. Remote sandbox gives OnePagent an optional isolated runtime for code-heavy work while keeping the app itself a single static HTML file.
  91. ---
  92. ## Deploy
  93. OnePagent is a pure static site — runs on any static host:
  94. <div align="center">
  95. [![Deploy with Vercel](https://vercel.com/button)](https://vercel.com/new/clone?repository-url=https%3A%2F%2Fgithub.com%2Fsligter%2FOnePagent&project-name=onepagent&repository-name=onepagent)
  96. &nbsp;
  97. [![Deploy on Zeabur](https://zeabur.com/button.svg)](https://zeabur.com/new)
  98. &nbsp;
  99. [![Deploy to Cloudflare Pages](https://img.shields.io/badge/Deploy-Cloudflare%20Pages-F38020?style=for-the-badge&logo=cloudflare&logoColor=white)](https://dash.cloudflare.com/?to=/:account/pages/new)
  100. </div>
  101. | Platform | Steps |
  102. |---|---|
  103. | **Vercel** | Click button, sign in, Deploy — done in ~10 seconds |
  104. | **Zeabur** | Click button, select your fork, auto-detected as static, Generate Domain |
  105. | **Cloudflare Pages** | Connect to Git, Framework: None, Build command: empty, Output: `/`, Deploy |
  106. | **GitHub Pages** | Settings, Pages, Source: GitHub Actions, push to deploy |
  107. > All data lives in the browser. Different domains do not share state. Use [Cloud Sync](#cloud-sync) for cross-domain / cross-device sync.
  108. ---
  109. ## Cloud Sync
  110. Back up and sync across devices via any S3-compatible bucket. Configure it in **Settings → Cloud Sync**.
  111. - Supports AWS S3, Cloudflare R2, Backblaze B2, and MinIO.
  112. - Required fields: Endpoint, Region, Bucket, Access Key ID, and Secret Access Key.
  113. - Optional: Prefix, AES-256-GCM encryption passphrase, auto-push, and path-style URLs for MinIO / R2.
  114. - Use **Test connection** / **Show CORS config** in Settings, then **Sync → Push now / Pull now** from the top bar.
  115. - LLM keys stay device-local and are never synced.
  116. ---
  117. ## Ralph Loop
  118. Ralph Loop keeps the agent working after a normal response ends. Turn on **Ralph**, send a task, and OnePagent will re-enter it until it sees the completion marker or a guard stops the run.
  119. - Configure defaults in **Settings → Ralph Loop**: max iterations, **Unlimited**, and completion marker (default `RALPH_DONE`).
  120. - Use the top-bar **Stop** button to cancel a run.
  121. - `AskUser` uses defaults or cancels instead of opening a modal; Plan Mode is never auto-approved.
  122. Example: `Audit this page, fix what you can, and when finished include RALPH_DONE.`
  123. ---
  124. ## Memory
  125. Long-term memory stores reusable facts across conversations. Enable it in **Settings → Memory**.
  126. - Stores `fact` / `preference` / `event` / `skill` / `note` records in IndexedDB.
  127. - Optional auto-extraction keeps durable facts after assistant turns.
  128. - Recall uses recency, tags, keywords, and prompt caching.
  129. - Tools: `memory_save`, `memory_search`, `memory_update`, `memory_forget`.
  130. - Memory Viewer supports search, filters, JSON import/export, and retired records.
  131. ---
  132. ## Agent Workflow
  133. - **Sub-agents**: the main agent can spawn bounded read-only research workers; runs and previews appear in the side panel.
  134. - **Human-in-the-loop**: `AskUser` supports text, choices, and confirmations for decisions the model should not guess.
  135. - **Media tools**: image/video generation is explicit through tools and configured generation models, not automatic after every turn.
  136. - **Diagnostics**: file-system diagnostics live in **Settings → Diagnostics**.
  137. ---
  138. ## Agent Swarm
  139. Opt-in parallel multi-agent runtime. Enable in **Settings → Agent Swarm**.
  140. - **Fanout** — lead emits several `SwarmSpawn(role, task)` calls in one turn; they run in parallel up to *Max concurrency*. Built-in roles: researcher, critic, writer, coder.
  141. - **Handoff** — workers chain via `SwarmHandoff(role, brief)`, e.g. `researcher → critic → writer`. Cycles and depth-overflow rejected.
  142. - **Blackboard** — per-turn shared workspace: `bb_write` / `bb_read` / `bb_list` / `bb_post_task` / `bb_claim`. Latest entries auto-injected into each worker's prompt.
  143. - **Custom roles** — manage in the left **Swarms** panel: own system prompt, tool whitelist, handoff targets, budgets. `bindSkills` grants any installed skill's tools. JSON import / export.
  144. - **RoleManager** *(opt-in)* — lead-only tool that creates / updates / deletes / duplicates roles at runtime. Storage choice: memory, global, or per-conversation. Built-in and user-authored roles are protected.
  145. Guards: per-worker and per-turn token budgets, concurrency cap, Plan Mode inheritance, Ralph Loop mutex, recursion blocked. Hooks `pre_swarm_spawn` / `post_swarm_spawn` join the existing six. Set *Worker model override* (e.g. Haiku) to keep workers cheap while the lead runs on a frontier model.
  146. Best for breadth-first work (research, comparison). Skip for tightly coupled refactors.
  147. ---
  148. ## Configuration
  149. All settings and conversations live in the browser (`localStorage` + `IndexedDB`). OnePagent has no server; requests go from your browser to your configured API endpoint.
  150. ---
  151. ## Contributing
  152. Single-file project — fork it, edit it, send a PR.
  153. - Keep `onepagent.html` runnable on its own
  154. - Avoid dependencies that require a build step
  155. - `index.html` is a redirect shim only — no real logic
  156. ---
  157. ## Star History
  158. <div align="center">
  159. [![Star History Chart](https://api.star-history.com/svg?repos=sligter/OnePagent&type=Date)](https://star-history.com/#sligter/OnePagent&Date)
  160. </div>
  161. ---
  162. ## Friends
  163. - [Linux Do](https://linux.do/) — A community for developers, by developers.
  164. ---
  165. ## License
  166. MIT (c) OnePagent contributors
  167. ---
  168. <div align="center">
  169. **One Page. Omnipotent Agent.**
  170. Built for people who believe a single HTML can still do everything.
  171. </div>

Ratings and comments

Sign in to rate this skill.

Sign in

Free to use. If it’s helped, pass it on.