Skip to main content
Glama

retrieve-codebase-insights

Analyze code repositories to extract insights on structure, dependencies, and patterns for informed development decisions.

Input Schema

TableJSON Schema
NameRequiredDescriptionDefault
repositoryUrlYes
insightTypesNo
tagsNo
relatedFileNo
limitNo

Implementation Reference

  • Registration and handler implementation for the 'retrieve-codebase-insights' MCP tool. Includes Zod input schema validation and logic to call the retrieveMemories helper, format results as JSON, and handle errors.
    server.tool( "retrieve-codebase-insights", { repositoryUrl: z.string(), insightTypes: z.array(z.string()).optional(), tags: z.array(z.string()).optional(), relatedFile: z.string().optional(), limit: z.number().optional() }, async ({ repositoryUrl, insightTypes, tags, relatedFile, limit }) => { try { const memories = await retrieveMemories({ repositoryUrl, insightTypes, tags, relatedFile, limit: limit || 10 }); return { content: [{ type: "text", text: JSON.stringify(memories, null, 2) }] }; } catch (error) { return { content: [{ type: "text", text: `Error retrieving insights: ${(error as Error).message}` }], isError: true }; } } );
  • Supporting helper function that implements the database query logic for retrieving codebase insights based on repository, types, tags, related files, and limit. Fetches base insights then populates related files and tags for each.
    export async function retrieveMemories(query: MemoryQuery): Promise<Insight[]> { if (!db) { db = await createDatabase("memory"); } if (!db) { throw new Error("Database not initialized"); } const { repositoryUrl, insightTypes, tags, relatedFile, limit } = query; // Base query let sql = ` SELECT i.id, i.repositoryUrl, i.insightType, i.category, i.insightContent, i.timestamp FROM insights i `; const params: any[] = [repositoryUrl]; // Add conditions for related files if (relatedFile) { sql += ` JOIN relatedFiles rf ON i.id = rf.insightId WHERE i.repositoryUrl = ? AND rf.filePath = ? `; params.push(relatedFile); } else { sql += ` WHERE i.repositoryUrl = ?`; } // Add conditions for insight types if (insightTypes && insightTypes.length > 0) { sql += ` AND i.insightType IN (${insightTypes.map(() => '?').join(',')})`; params.push(...insightTypes); } // Add conditions for tags if (tags && tags.length > 0) { sql += ` AND i.id IN ( SELECT insightId FROM tags WHERE tag IN (${tags.map(() => '?').join(',')}) GROUP BY insightId HAVING COUNT(DISTINCT tag) = ? ) `; params.push(...tags, tags.length); // Ensure all tags match } // Add order and limit sql += ` ORDER BY i.timestamp DESC`; if (limit) { sql += ` LIMIT ?`; params.push(limit); } // Execute query const insights = await db.all(sql, params); // For each insight, fetch related files and tags for (const insight of insights) { // Get related files insight.relatedFiles = await db.all( `SELECT filePath FROM relatedFiles WHERE insightId = ?`, [insight.id] ).then(rows => rows.map((row: any) => row.filePath)); // Get tags insight.tags = await db.all( `SELECT tag FROM tags WHERE insightId = ?`, [insight.id] ).then(rows => rows.map((row: any) => row.tag)); } return insights; }

Latest Blog Posts

MCP directory API

We provide all the information about MCP servers via our MCP API.

curl -X GET 'https://glama.ai/api/mcp/v1/servers/0xjcf/MCP_CodeAnalysis'

If you have feedback or need assistance with the MCP directory API, please join our Discord server