ToolFront MCP Server

--- icon: lucide/package-open --- # Build your first app Build your first RAG application with vector search, text-to-SQL, and agentic RAG. ## Define project layout Start with this directory structure: ```bash project/ ├── README.md ├── src/ │ ├── agentic_rag.md │ ├── text2sql.md │ └── vector_search.md └── data/ ├── log1.txt ├── log2.txt └── log3.txt 2 directories, 7 files ``` ## Set up README Create the `README.md` with tools to navigate other files: ```yaml title="README.md" --- tools: - [ls] - [cat] --- # My first application - Use `ls` and `cat` to navigate other files ``` ## Add vector search Create a page for querying vector embeddings: ```yaml title="src/vector_search.md" --- tools: - [curl, -X, POST, https://host.pinecone.io/records/namespaces/user/search] --- # Vector search instructions: - Query documents with your vector database API ``` > Replace the URL with your vector database API endpoint (Pinecone, Weaviate, Qdrant, etc.). ## Add text-to-SQL Create a page for read-only database queries: ```yaml title="src/text2sql.md" --- tools: - [psql, -U, $USER, -d, $DB, -c, { regex: "^SELECT\b.*" }] --- # Text-to-SQL Instructions: - Use `psql` for read-only PostgreSQL queries ``` > Alternatively, use your own database CLI (mysql, sqlite3, mongosh, etc.). ## Add agentic RAG Add sample log files to `data/`: ```text title="data/doc1.txt" 2025-01-15 14:23:01 INFO [api.health] GET /status/200 - 200 OK (12ms) 2025-01-15 14:23:45 INFO [api.health] Health check passed for service-1 2025-01-15 14:24:12 INFO [auth] User 'user1' authenticated successfully ``` ```text title="data/doc2.txt" 2025-02-03 09:15:22 ERROR [api.handler] GET /api/users - 500 Internal Server Error 2025-02-03 09:15:22 ERROR [database] Connection timeout after 5000ms 2025-02-03 09:15:23 WARN [api.handler] Retrying failed request (attempt 1/3) ``` ```text title="data/doc3.txt" 2025-02-15 16:42:10 INFO [worker] Job 'daily-backup' started 2025-02-15 16:42:11 INFO [worker] Processing 1250 records 2025-02-15 16:42:33 INFO [worker] Job 'daily-backup' completed (status: success) ``` Then, create a page for searching through the logs with `grep`: ```yaml title="src/agentic_rag.md" --- tools: - [grep, -r, -i, { }, ../data/] --- # Document search instructions: - Use `grep` to search documents in `../data/`. ``` > The `{ }` placeholder requires agents to pass exactly one argument. --- !!! success You've built your first RAG application. **Next**: [deploy your app](deploy.md)