AQBot 0.0.164

? Lightweight, high-performance cross-platform AI chat + AI gateway desktop client

Features
Chat and Models
- Multi-provider support -- Compatible with all OpenAI-compatible APIs including OpenAI, Anthropic Claude, Google Gemini, etc.
- Model management -- Supports remote fetching of model lists and custom parameters (temperature, max tokens, top-p, etc.)
- Multi-key round-robin -- Configure multiple API keys per provider; automatically rotates to distribute rate-limit pressure
- Streaming output -- Real-time token-by-token rendering; thinking blocks can be folded or expanded
- Message versions -- Each response supports multiple versions for easy comparison of different models or parameter settings
- Conversation branching -- Create new branches from any message node, enabling side-by-side branch comparisons
- Conversation management -- Supports pinning, archiving, time-based grouping, and batch operations
- Conversation compression -- Automatically compresses long conversations, preserving key information to save context space
- Multi-model same query -- Ask the same question to multiple models simultaneously, with support for comparative analysis of responses
AI Agent
- Agent mode -- Switch to Agent mode, allowing AI to autonomously perform multi-step tasks: reading/writing files, running commands, code analysis, etc.
- Three-tier permissions -- Default mode (edits require approval), accept edits (automatically approve changes), full access (no prompts), ensuring secure and controlled operation
- Workspace sandbox -- Agent operations strictly confined to designated workspace directory, preventing unauthorized access
- Tool approval panel -- Real-time display of tool invocation requests, supporting per-item review, one-click always allow or deny
- Cost tracking -- Real-time token usage and cost statistics for each session
Content Rendering
- Markdown rendering -- Full support for syntax highlighting, LaTeX math formulas, tables, and task lists
- Monaco code editor -- Embedded Monaco Editor in code blocks, supporting syntax highlighting, copy, and diff preview
- Chart rendering -- Built-in Mermaid flowchart and D2 architecture diagram rendering
- Artifact panel -- Preview code snippets, HTML drafts, Markdown notes, reports in isolated panels
- Real-time voice chat -- (Coming soon) WebRTC-based real-time voice, compatible with OpenAI Realtime API
Search and Knowledge
- Web search -- Integrated with Tavily, Zhipu WebSearch, Bocha, etc., with search results annotated with source references
- Local knowledge base (RAG) -- Supports multiple knowledge bases; upload documents to automatically parse, segment, and build indexes, then semantically retrieve relevant passages during chat
- Memory system -- Supports multi-namespace conversation memory; entries can be manually added or automatically extracted by AI (AI auto-extraction coming soon)
- Context management -- Flexibly attach files, search results, knowledge base snippets, memory entries, and tool outputs
Tools and Extensions
- MCP protocol -- Fully implements Model Context Protocol, supporting both stdio and HTTP transport methods
- Built-in tools -- Includes ready-to-use built-in MCP tools such as @aqbot/fetch
- Tool execution panel -- Visual display of tool invocation requests and return results
API Gateway
- Local API gateway -- Built-in local API server with native OpenAI-compatible, Claude, Gemini interfaces; can serve as backend for any compatible client
- API key management -- Generate, revoke, enable, or disable access keys with descriptive notes
- Usage statistics -- Analyze request volume and token usage by key, provider, and date
- SSL/TLS support -- Built-in self-signed certificate generation, also supports mounting custom certificates
- Request logs -- Fully record all API requests and responses passing through the gateway
- Configuration templates -- Pre-configured integration templates for common CLI tools like Claude, Codex, OpenCode, Gemini
Data and Security
- AES-256 encryption --...

License: AGPL-3.0