---

name: data-sql-graph-agents description: "Use Langroid structured-data agents for table/CSV/Pandas, SQL, Neo4j, ArangoDB, and CSV-to-knowledge-graph workflows with schema/context setup and safe query validation." disable-model-invocation: true metadata: disco-role: operating license: MIT

Data, SQL, and Graph Agents

Use this sub-skill for structured data questions, tabular analysis, SQL-backed Q&A, and graph construction or graph Q&A.

Choose the right agent

• Table, CSV, or Pandas analysis -> TableChatAgent
• SQL database Q&A -> SQLChatAgent
• Neo4j graph Q&A or writes -> Neo4jChatAgent
• ArangoDB graph Q&A or writes -> ArangoChatAgent
• CSV-to-knowledge-graph bootstrapping -> CSVGraphAgent

Route elsewhere

• Unstructured document RAG -> sibling skill retrieval-doc-chat
• Provider, model, or API-key setup -> sibling skill llm-provider-config
• Generic Task / Tool orchestration -> sibling skill agents-tasks-tools

Start here

• API reference
• Workflow patterns
• Database and schema config
• Security and validation
• Troubleshooting
• Safety smoke script

Operating rules

• Prefer the least-privilege, validation-first path.
• Keep TableChatAgentConfig.full_eval=False unless the input is fully trusted.
• Keep SQLChatAgentConfig.allow_dangerous_operations=False and extend allowed_statement_types only when writes are intended.
• Keep Neo4j and Arango dangerous operations disabled by default.
• Use schema or context helpers before asking the model to write SQL or graph queries.
• Use the bundled smoke script to inspect defaults or validate a SQL, Cypher, or AQL string locally without touching a live service.
• If the question is really about model/provider configuration, hand it off to the provider-config skill instead of forcing it through this one.