m9k_context

Instructions

Implementation Reference

• src/tools/search.ts:69-127 (handler)

  Registration and handler for the m9k_context tool. Accepts chunkId and optional window parameter. Queries the conv_chunks table for the target chunk, then retrieves adjacent chunks in the same session to provide surrounding conversation context.

  server.registerTool(
    'm9k_context',
    {
      description:
        'Get a chunk with surrounding context (adjacent chunks in the same session). Use after m9k_search to understand the conversation flow.',
      inputSchema: {
        chunkId: z.string().describe('The chunk ID to get context for'),
        window: z
          .number()
          .int()
          .min(1)
          .max(10)
          .default(3)
          .describe('Number of chunks before/after to include'),
      },
      annotations: {
        readOnlyHint: true,
        destructiveHint: false,
        idempotentHint: true,
        openWorldHint: false,
      },
    },
    async ({ chunkId, window }) => {
      const target = ctx.db
        .prepare('SELECT session_id, idx FROM conv_chunks WHERE id = ? AND deleted_at IS NULL')
        .get(chunkId) as { session_id: string; idx: number } | undefined;
  
      if (!target) {
        return {
          content: [
            {
              type: 'text' as const,
              text: JSON.stringify({ error: 'Chunk not found', chunkId }),
            },
          ],
          isError: true,
        };
      }
  
      const minIdx = Math.max(0, target.idx - window);
      const maxIdx = target.idx + window;
  
      const rows = ctx.db
        .prepare(
          `SELECT id, session_id, idx, kind, user_content, assistant_content, hash, timestamp, token_count, tags, metadata_json
           FROM conv_chunks WHERE session_id = ? AND idx BETWEEN ? AND ? AND deleted_at IS NULL ORDER BY idx`,
        )
        .all(target.session_id, minIdx, maxIdx);
  
      return {
        content: [
          {
            type: 'text' as const,
            text: JSON.stringify({ target: chunkId, chunks: rows }),
          },
        ],
      };
    },
  );

• src/tools/search.ts:69-89 (schema)

  Input schema for m9k_context: chunkId (string, required) and window (number, 1-10, default 3) defining how many chunks before/after to include.

  server.registerTool(
    'm9k_context',
    {
      description:
        'Get a chunk with surrounding context (adjacent chunks in the same session). Use after m9k_search to understand the conversation flow.',
      inputSchema: {
        chunkId: z.string().describe('The chunk ID to get context for'),
        window: z
          .number()
          .int()
          .min(1)
          .max(10)
          .default(3)
          .describe('Number of chunks before/after to include'),
      },
      annotations: {
        readOnlyHint: true,
        destructiveHint: false,
        idempotentHint: true,
        openWorldHint: false,
      },

• src/tools/search.ts:69-127 (registration)

  Tool registration via server.registerTool('m9k_context', ...) with schema, annotations (readOnly, idempotent), and async handler. Exported by registerSearchTools function.

  server.registerTool(
    'm9k_context',
    {
      description:
        'Get a chunk with surrounding context (adjacent chunks in the same session). Use after m9k_search to understand the conversation flow.',
      inputSchema: {
        chunkId: z.string().describe('The chunk ID to get context for'),
        window: z
          .number()
          .int()
          .min(1)
          .max(10)
          .default(3)
          .describe('Number of chunks before/after to include'),
      },
      annotations: {
        readOnlyHint: true,
        destructiveHint: false,
        idempotentHint: true,
        openWorldHint: false,
      },
    },
    async ({ chunkId, window }) => {
      const target = ctx.db
        .prepare('SELECT session_id, idx FROM conv_chunks WHERE id = ? AND deleted_at IS NULL')
        .get(chunkId) as { session_id: string; idx: number } | undefined;
  
      if (!target) {
        return {
          content: [
            {
              type: 'text' as const,
              text: JSON.stringify({ error: 'Chunk not found', chunkId }),
            },
          ],
          isError: true,
        };
      }
  
      const minIdx = Math.max(0, target.idx - window);
      const maxIdx = target.idx + window;
  
      const rows = ctx.db
        .prepare(
          `SELECT id, session_id, idx, kind, user_content, assistant_content, hash, timestamp, token_count, tags, metadata_json
           FROM conv_chunks WHERE session_id = ? AND idx BETWEEN ? AND ? AND deleted_at IS NULL ORDER BY idx`,
        )
        .all(target.session_id, minIdx, maxIdx);
  
      return {
        content: [
          {
            type: 'text' as const,
            text: JSON.stringify({ target: chunkId, chunks: rows }),
          },
        ],
      };
    },
  );

• src/tools/search.ts:1-9 (helper)

  Imports and setup for the search module: zod for schema validation, McpServer type, getStat for DB stats, search function, and ToolContext interface.

  /**
   * Progressive retrieval tools: m9k_search, m9k_context, m9k_full.
   */
  
  import { z } from 'zod';
  import type { McpServer } from '@modelcontextprotocol/sdk/server/mcp.js';
  import { getStat } from '../db.js';
  import { search } from '../search.js';
  import type { ToolContext } from './context.js';