feat(context): Phase 7 - Main Engine & Integration #85

New Issue

cardosofelipe · 2026-01-04T00:52:21Z

cardosofelipe commented

2026-01-04 00:52:21 +00:00

Overview

Implement the main ContextEngine class and integrate with MCPClientManager.

Parent Issue

#61: Context Management Engine

Implementation Tasks

1. Create `engine.py`

Create ContextEngine class
Initialize all components (calculator, scorer, ranker, compressor, formatter, cache)
Implement assemble_context() main method
Implement _fetch_knowledge() for RAG integration
Implement _convert_history() for conversation context

class ContextEngine:
    """Main context management engine."""

    def __init__(
        self,
        mcp_manager: MCPClientManager,
        redis: Redis | None,
        settings: ContextSettings | None = None
    ):
        self.mcp = mcp_manager
        self.settings = settings or ContextSettings()

        # Initialize components
        self.calculator = TokenCalculator(mcp_manager)
        self.scorer = CompositeScorer()
        self.ranker = ContextRanker()
        self.compressor = Compressor()
        self.formatter = ContextFormatter()
        self.cache = ContextCache(redis, self.settings) if redis else None

        self.pipeline = ContextPipeline(...)

    async def assemble_context(
        self,
        project_id: str,
        agent_id: str,
        query: str,
        model: str,
        max_tokens: int,
        system_prompt: str | None = None,
        conversation_history: list[dict] | None = None,
        knowledge_query: str | None = None,
        task_description: str | None = None
    ) -> AssembledContext:
        """Assemble optimized context for an LLM request."""
        ...

2. Create FastAPI Dependency

Create get_context_engine() dependency
Handle Redis optionality
Implement singleton pattern

# backend/app/api/dependencies/context.py
_engine: ContextEngine | None = None
_lock = asyncio.Lock()

async def get_context_engine(
    mcp: MCPClientManager = Depends(get_mcp_manager),
    redis: Redis = Depends(get_redis)
) -> ContextEngine:
    global _engine
    if _engine is None:
        async with _lock:
            if _engine is None:
                _engine = ContextEngine(mcp, redis)
    return _engine

3. Update `init.py`

Export ContextEngine as main public API
Export AssembledContext
Export ContextSettings

4. Integration with MCPClientManager

Add context_engine property to MCPClientManager
Ensure proper initialization order

Files to Create/Modify

backend/app/services/context/
├── engine.py                  # NEW
└── __init__.py                # UPDATE

backend/app/api/dependencies/
└── context.py                 # NEW

Acceptance Criteria

ContextEngine assembles context correctly
Knowledge Base integration works (RAG)
Conversation history is converted properly
FastAPI dependency provides singleton
Caching is enabled when Redis available
Integration tests pass

Dependencies

#79 (Phase 1 - Foundation)
#80 (Phase 2 - Token Budget)
#81 (Phase 3 - Scoring & Ranking)
#82 (Phase 4 - Assembly Pipeline)
#83 (Phase 5 - Model Adapters)
#84 (Phase 6 - Caching Layer)

Labels

phase-2, context, backend

## Overview Implement the main ContextEngine class and integrate with MCPClientManager. ## Parent Issue - #61: Context Management Engine --- ## Implementation Tasks ### 1. Create `engine.py` - [ ] Create `ContextEngine` class - [ ] Initialize all components (calculator, scorer, ranker, compressor, formatter, cache) - [ ] Implement `assemble_context()` main method - [ ] Implement `_fetch_knowledge()` for RAG integration - [ ] Implement `_convert_history()` for conversation context ```python class ContextEngine: """Main context management engine.""" def __init__( self, mcp_manager: MCPClientManager, redis: Redis | None, settings: ContextSettings | None = None ): self.mcp = mcp_manager self.settings = settings or ContextSettings() # Initialize components self.calculator = TokenCalculator(mcp_manager) self.scorer = CompositeScorer() self.ranker = ContextRanker() self.compressor = Compressor() self.formatter = ContextFormatter() self.cache = ContextCache(redis, self.settings) if redis else None self.pipeline = ContextPipeline(...) async def assemble_context( self, project_id: str, agent_id: str, query: str, model: str, max_tokens: int, system_prompt: str | None = None, conversation_history: list[dict] | None = None, knowledge_query: str | None = None, task_description: str | None = None ) -> AssembledContext: """Assemble optimized context for an LLM request.""" ... ``` ### 2. Create FastAPI Dependency - [ ] Create `get_context_engine()` dependency - [ ] Handle Redis optionality - [ ] Implement singleton pattern ```python # backend/app/api/dependencies/context.py _engine: ContextEngine | None = None _lock = asyncio.Lock() async def get_context_engine( mcp: MCPClientManager = Depends(get_mcp_manager), redis: Redis = Depends(get_redis) ) -> ContextEngine: global _engine if _engine is None: async with _lock: if _engine is None: _engine = ContextEngine(mcp, redis) return _engine ``` ### 3. Update `__init__.py` - [ ] Export ContextEngine as main public API - [ ] Export AssembledContext - [ ] Export ContextSettings ### 4. Integration with MCPClientManager - [ ] Add `context_engine` property to MCPClientManager - [ ] Ensure proper initialization order --- ## Files to Create/Modify ``` backend/app/services/context/ ├── engine.py # NEW └── __init__.py # UPDATE backend/app/api/dependencies/ └── context.py # NEW ``` --- ## Acceptance Criteria - [ ] ContextEngine assembles context correctly - [ ] Knowledge Base integration works (RAG) - [ ] Conversation history is converted properly - [ ] FastAPI dependency provides singleton - [ ] Caching is enabled when Redis available - [ ] Integration tests pass --- ## Dependencies - #79 (Phase 1 - Foundation) - #80 (Phase 2 - Token Budget) - #81 (Phase 3 - Scoring & Ranking) - #82 (Phase 4 - Assembly Pipeline) - #83 (Phase 5 - Model Adapters) - #84 (Phase 6 - Caching Layer) ## Labels `phase-2`, `context`, `backend`

cardosofelipe referenced this issue

2026-01-04 00:54:11 +00:00

feat(mcp): Context Management Engine #61

cardosofelipe referenced this issue from a commit

2026-01-04 14:23:32 +00:00

feat(context): implement main ContextEngine with full integration (#85)

cardosofelipe closed this issue

2026-01-04 18:53:45 +00:00

Sign in to join this conversation.