cyper-rag-server

"""
tools/ingest_tool.py — MCP tool to trigger live re-ingestion at runtime.

This lets the LLM (or an operator) add new documents to the docs folder and
re-index without restarting the server.
"""

from __future__ import annotations

from mcp import types

from src.retrieval.engine import RAGEngine
from src.utils.logging import get_logger

log = get_logger("tools.ingest")


def build_ingest_tool_schema() -> types.Tool:
    return types.Tool(
        name="reindex_documents",
        description=(
            "Re-scan the documents folder and index any new or changed files. "
            "Use this after adding new documents to the watched folder. "
            "Already-indexed unchanged files are skipped automatically."
        ),
        inputSchema={
            "type": "object",
            "properties": {},
            "required": [],
        },
    )


async def handle_reindex(engine: RAGEngine) -> list[types.TextContent]:
    """Trigger a live re-ingest and return a summary."""
    log.info("Live re-ingest triggered via MCP tool.")
    try:
        stats = engine.ingest()
        msg = (
            f"Re-indexing complete.\n"
            f"• New files indexed : {stats['new']}\n"
            f"• Unchanged skipped : {stats['skipped']}\n"
            f"• Total chunks      : {stats['total_chunks']}"
        )
    except Exception as exc:
        log.error("Re-ingest failed: %s", exc)
        msg = f"Re-indexing failed: {exc}"

    return [types.TextContent(type="text", text=msg)]