[WIP] Update roadmap for version 0.1 development

[Copilot]

• Analyze the Roadmap v0.1 issue and compare with r-dev branch code
• Create src/chat/utils/common_utils.py with TempMethodsHFC class
• Remove methods _parse_stream_config_to_chat_id, _now_minutes, _parse_range, _in_range, get_talk_value from ChatConfig in official_configs.py
• Update heartFC_chat.py to use TempMethodsHFC.get_talk_value()
• Update frequency_api.py to use TempMethodsHFC.get_talk_value()

Original prompt

---

This section details on the original issue you should resolve

<issue_title>Roadmap v0.1</issue_title>
<issue_description>> Note: All development work mentioned in this roadmap is being actively worked on in the r-dev branch.

What have we done previously?

MaiBot has evolved from a simple bot extension to a full-featured AI agent. Here's the version history:

• v0.12.x (2025-12 ~ 2026-01) - Thinking intensity mechanism, dream functionality, MCP plugin, WebUI enhancements
• v0.11.x (2025-10 ~ 2025-12) - First official WebUI, memory system (Hippocampus Agent), jargon learning
• v0.10.x (2025-08 ~ 2025-09) - Major refactoring: tool system merged into plugins, LLM request refactoring
• v0.9.x (2025-07) - Plugin system overhaul, Normal/Focus mode unification
• v0.8.x (2025-06 ~ 2025-07) - Plugin system launch, character impression system
• v0.7.x (2025-06) - MongoDB → SQLite migration, expression learning
• v0.6.x (2025-04 ~ 2025-05) - MaiCore refactor, thinking flow system, LPMM knowledge base

---

v0.1 Features & Roadmap

🚀 Top Priorities (Mai NEXT)

🏗️ Architecture Refactoring

• Configuration System Overhaul

  • TOML configuration format
  • Pydantic-based config classes with atomic types only
  • Migrate methods to temporary classes
    • TempMethodsExpression (migrated from ExpressionConfig)
    • TempMethodsLLMUtils (migrated from ModelConfig)
    • TempMethodsHFC (from ChatConfig) — class not yet created; _parse_range / get_talk_value still on HeartFChatting
  • BotConfig simplification (Nickname & Alias Name only)
  • Configuration hot-reload system
    • File watcher (watchfiles)
    • Callback registration with error isolation
    • Lock mechanism for safe reloading (asyncio.Lock)

• Database Model Unification

  • Auto-increment primary key id for all tables
  • SQLModel as base class
  • JSON storage for complex types
  • Unified timestamp field naming — currently 8+ patterns (timestamp, record_time, create_time, created_timestamp, first_known_time, etc.)
  • Soft delete mechanism for images/emojis (is_deleted field)
  • Unified naming: stream_id/chat_id → session_id — SQLModel tables mostly migrated; legacy data models still use old names
  • Hash algorithm upgrade to SHA-256 — partially done (images, emoji, knowledge graph use SHA-256; session ID, person info, message utils still use MD5)

• Message System

  • Detailed message construction documentation
    • Mixed type documentation
    • Forward type documentation
    • Command type documentation
  • Message chain construction (AstrBot-style)
  • Accept Format validation before sending

🧠 Core Systems

• LLM Interface Unification

  • Unified LLM response format — APIResponse and LLMGenerationDataModel both exist; needs consolidation into a single model
  • Remove LLM Client initialization from __init__ — now uses ClientRegistry pattern
  • get_llm_client() method for all LLM access — currently uses client_registry.get_client_class_instance()
  • LLM Client Manager (ClientRegistry partially implements this)
    • Singleton/multi-instance management
    • Cache management & lifecycle
    • Hot-reload on config change

• Memory System Enhancement

  • Time-aware memory system
    • Query/delete by timestamp
  • Memory layering
    • Instant memory
    • Short-term memory
    • Long-term memory
  • Integration with LPMM knowledge graph

• Prompt Management

  • User-defined Prompt system
  • Prompt construction & file interaction
  • Official Prompts independence

🔌 Plugin System

• Plugin Manager Overhaul

  • Rewrite plugin loading logic
  • Subclass-based loading (drop @register_plugin)
  • Plugin hot-reload system
    • Config file hot-reload
    • Code hot-reload (with rollback support)
  • allow_hot_reload: bool flag for plugins

• Component Manager

  • Better component management code
  • Global/local control parity
  • Component property modification API
  • Component system unloading

• Dependency Management

  • requirements.txt analysis
  • python_dependencies auto-install
  • plugin_dependencies with topological sort

• MCP Integration — MCP Bridge Plugin already exists (plugins/MaiBot_MCPBridgePlugin/) with tool chain, circuit breaker, and WebUI support; items below are for core-level adapter design

  • MCP adapter class design
    • MCP call prompt construction
    • Content passing
    • Result handling

• Tool Result Caching

  • put_cache() method
  • get_cache() method
  • need_cache flag

• **Events Dependency Me...

• Fixes Roadmap v1.0 #1531

---

💬 We'd love your input! Share your thoughts on Copilot coding agent in our 2 minute survey.