🤖 AicqWebBot

Your browser is the agent's container.

Your browser is the agent's container. — 中文说明

AicqWebBot is a browser-native AI agent runtime. pip install it, run one line, open localhost, paste your model key — and a full agent (LLM loop, 50+ tools, WASM code sandbox, memory) starts running entirely inside your browser tab.

The local server it starts is a thin, stateless relay: it only proxies your model API calls (to bypass browser CORS) and serves static files. Zero execution happens on the server. Your API key, your conversation history, your files — everything stays on your machine.

🚀 Quick start (3 lines)

pip install aicqwebbot

import aicqwebbot
aicqwebbot.run(8386)

Open http://localhost:8386 → paste
your OpenAI-compatible key → chat.

Want it even simpler? Try the live demo spaces — no install at all:

🤗 Hugging Face: https://huggingface.co/spaces/samaiccgroup/AicqWebBot
🧲 ModelScope: https://www.modelscope.cn/studios/samaiccgroup/AicqWebBot

🧠 Why this is different

	Traditional agent stacks	AicqWebBot
Agent loop (brain)	Server container	Your browser tab
Code execution (hands)	Sandbox containers (E2B/Modal/...)	WASM sandbox in your browser (Pyodide / QuickJS)
Memory & files	Database container	IndexedDB on your device
Model calls	LLM gateway container	Local relay or direct — your key, your machine
Scaling cost	Grows with users × execution time	Zero — every user brings their own compute
Blast radius	Untrusted code runs on YOUR server	Untrusted code runs on the USER's own browser — the best-tested sandbox on Earth (Chromium)

The insight: containers are scarce on servers, but the browser is a free, pre-installed, perfectly isolated container on every machine on the planet. AicqWebBot simply stops paying rent on server containers that the browser already replaces.

🛠 What's in the box

The bundled agent runtime (a zero-modification build of the aicq.me web agent) ships with:

50+ built-in tools — web search, web read, HTTP client, file system (virtual FS), doc/xlsx/pdf/ppt creation, charts, image drawing, QR codes, email, clipboard, translate, weather, task planning, and more
Free anonymous models — no API key at all: OpenCode Zen free models (0.4.10+). Calls egress from your machine via the local relay — never from any server
Code sandbox — Python (Pyodide / numpy / pandas) or JavaScript (QuickJS), executing as WebAssembly inside your tab
Context compression — long conversations are automatically summarized so memory usage stays bounded
Vision — send images, the agent analyzes them with a vision model
Streaming UI — live reasoning, tool-call cards, markdown rendering

🔧 Architecture

pip install aicqwebbot && aicqwebbot.run(8386)
        │
        ▼
┌─ localhost:8386 (thin Python relay) ─────────────────┐
│  GET  /                      config page / chat UI   │
│  GET  /static/*              agent runtime bundle    │
│  POST /api/v1/agent/llm-proxy     -> your LLM API    │
│  POST /api/v1/agent/search-proxy  -> web search      │
│  POST /api/v1/agent/web-proxy     -> page fetcher    │
└──────────────────────────────────────────────────────┘
        │
        ▼
  Your browser tab = the agent's container
  ┌───────────────────────────────────────────────┐
  │  agent loop · 50 tools · WASM sandbox         │
  │  memory (IndexedDB) · virtual file system     │
  └───────────────────────────────────────────────┘

🔐 Privacy model

Your API key lives in your browser's storage and is only sent as an Authorization header to your chosen model provider, relayed verbatim through the local proxy (in-memory only, never persisted or logged).
Agent memory, conversation history, and virtual files never leave your device.
The relay keeps no sessions, no accounts, no database. Delete the browser data and everything is gone — your machine, your data.

📦 Deployment options

Target	Command	Notes
Local machine	`aicqwebbot.run(8386)`	The default experience
Hugging Face Space	push `deploy/hf/`	Free, one-container demo
ModelScope Studio	push `deploy/modelscope/`	Same, for the CN community
Any server / intranet	`aicqwebbot.run(port)` behind nginx	Stateless — scale horizontally for free

🧩 Where the code comes from

aicqwebbot/web/static/agent/* is the exact browser agent runtime that powers aicq.me's local agents (agent-engine, tool registry, WASM sandboxes, IndexedDB storage). It is synchronized automatically from the private samaidev/aicq monorepo — fixes upstream flow downstream with every release.

📄 License

Apache-2.0

Metadata

Release files for aicqwebbot 0.4.17

For a detailed explanation of source distributions (sdists) and built distributions (wheels), please see the package formats documentation.

Source distribution (sdist)

Source distribution for aicqwebbot 0.4.17
File	Size	Uploaded	Provenance
aicqwebbot-0.4.17.tar.gz	223.4 kB	Oct 7, 2026	Without provenance	Details

Built distribution (wheel)

Table of built distributions (wheels) for aicqwebbot 0.4.17
File	Interpreter	ABI	Platform	Reset
aicqwebbot-0.4.17-py3-none-any.whl 237.0 kB Oct 7, 2026 Without provenance	Python 3	none	any	Details

Total release size: 460.4 kB

Release files / aicqwebbot-0.4.17.tar.gz

Download URL	aicqwebbot-0.4.17.tar.gz
Size	223.4 kB
Tags	Source
SHA-256 checksum How to use checksums	`a6b716fc2b1157ef010bbfaf2a343cbb1fa65e07f70569472bcdfc16ad682282`
BLAKE2b-256 checksum How to use checksums	`2e6108a9faf16fab7477d62ff6988adccd424a0a25cf043357d2f4ca4e4d2a12`
Upload date	Oct 7, 2026
Uploaded using Trusted Publishing? What is trusted publishing?	No
Uploaded via	`twine/7.0.0 CPython/3.14.4`

Release files / aicqwebbot-0.4.17-py3-none-any.whl

Download URL	aicqwebbot-0.4.17-py3-none-any.whl
Size	237.0 kB
Tags	Python 3
SHA-256 checksum How to use checksums	`8a823ecf4edbec554d35487882af92fa787d4fad394eab4cef26943b2a2034c3`
BLAKE2b-256 checksum How to use checksums	`98b451c06c59dc95ad78fcbf532219985818c0762f50f2db6e871f3e424e362a`
Upload date	Oct 7, 2026
Uploaded using Trusted Publishing? What is trusted publishing?	No
Uploaded via	`twine/7.0.0 CPython/3.14.4`

Security for aicqwebbot 0.4.17

Report project as malware

Provenance

Provenance describes where a file came from. On PyPI, provenance is shared via attestations, which provide a verifiable record of the build or publishing details. View details, limitations and caveats.

No provenance found

None of the files in this release were uploaded with provenance metadata.

A note on package security

PyPI is a repository for distributing packages, not a service for auditing them. While third-party security researchers and our staff do their best to detect and remove malicious packages, PyPI cannot guarantee the quality, security or suitability of any given package.

To learn more about protecting yourself from malicious actors, visit our package security documentation.