{"slug": "a-langgraph-sql-agent-that-only-sees-the-tables-it-needs", "title": "A LangGraph SQL agent that only sees the tables it needs", "summary": "A new LangGraph example in the open-source schemagate project cuts schema tokens sent to the model by 68% to 78% on a bundled 42-table demo database, sending 489 to 706 tokens per question instead of 2,199, with all three test answers correct using Claude Haiku. The four-node agent uses a SchemagateRetriever to pick only the tables a question needs in an offline step with no model call, then writes one SELECT, runs it, and returns a one-sentence answer; on databases with 50 or more tables the saving reaches about 91%. Schemagate, which is Apache-2.0 and listed in LangChain's official retriever integrations, ranks #1 on the BIRD Mini-Dev leaderboard across SQLite, PostgreSQL and MySQL, and the example installs via pip install langchain-schemagate langgraph langchain-anthropic tiktoken.", "body_md": "# A LangGraph SQL agent that only sees the tables it needs\n\nMost text-to-SQL agents paste the whole schema into every prompt. On a real database, most of it is tables the question never touches, and you pay for them on every call. I added a small LangGraph example to schemagate that does it differently. It has four nodes: select_schema: SchemagateRetriever p\n\nicks the few tables the question needs. This step runs offline, with no model call. write_sql: the model writes one SELECT using only those tables. run_sql: the agent runs the query.\n\nIf it fails, the error goes back to the model once. answer: the model answers in one sentence from the returned rows. On the bundled 42-table demo database: Question Schema tokens sent Which customers owe us money? 489 vs 2,199 (78% fewer) How many orders did each carrier ship?\n\n706 vs 2,199 (68% fewer) Which products are below their reorder level, and in which warehouse? 563 vs 2,199 (74% fewer) All three answers were correct with Claude Haiku. On databases with 50+ tables the saving reaches about 91%. The selection also respects roles: a table the caller can't read never appears in the prompt.\n\npip install langchain-schemagate langgraph langchain-anthropic tiktoken python examples/langgraph_sql_agent.py To use your own database, swap demo_catalog() for Catalog().bootstrap(\"postgresql://...\"). schemagate is #1 on the BIRD Mini-Dev leaderboard (SQLite, PostgreSQL, MySQL), is listed in LangChain's official retriever integrations, and is Apache-2.0. Example: examples/langgraph_sql_agent.py\n\n## Key Takeaways\n\n- •This step runs offline, with no model call.\n- •This story was reported by **Dev.to** , covering developments in the**dev** space.\n- •AI advancements continue to reshape industries — read the full article on Dev.to for complete coverage.\n\n📖 Continue reading the full article:\n\n[Read Full Article on Dev.to →](https://dev.to/ashish_sinha_5241c7673d93/a-langgraph-sql-agent-that-only-sees-the-tables-it-needs-23no)", "url": "https://wpnews.pro/news/a-langgraph-sql-agent-that-only-sees-the-tables-it-needs", "canonical_source": "https://ainexusdaily.vercel.app/article/2026-11-06-a-langgraph-sql-agent-that-only-sees-the-tables-it-needs", "published_at": "2026-11-06 04:15:07+00:00", "updated_at": "2026-11-06 05:50:47.360783+00:00", "lang": "en", "topics": ["ai-agents", "large-language-models", "ai-tools", "developer-tools", "ai-research"], "entities": ["LangGraph", "schemagate", "SchemagateRetriever", "Claude Haiku", "LangChain", "BIRD Mini-Dev", "Dev.to", "PostgreSQL"], "also_reported_by": [], "alternates": {"html": "https://wpnews.pro/news/a-langgraph-sql-agent-that-only-sees-the-tables-it-needs", "markdown": "https://wpnews.pro/news/a-langgraph-sql-agent-that-only-sees-the-tables-it-needs.md", "text": "https://wpnews.pro/news/a-langgraph-sql-agent-that-only-sees-the-tables-it-needs.txt", "jsonld": "https://wpnews.pro/news/a-langgraph-sql-agent-that-only-sees-the-tables-it-needs.jsonld"}}