/projects/boundeddata
Bounded Data Agent
受治理的检索增强数据智能体平台
/projects/boundeddata
受治理的检索增强数据智能体平台
Bounded Data Agent 处理的核心挑战,是把受治理的检索增强数据智能体平台变成有用、可治理且可解释的产品系统。
Bounded Data Agent 围绕明确边界、可检查工作流和可见交付技术栈组织,其中包括 TypeScript、Node.js、PostgreSQL和pgvector。
难点在于让自主性、数据、工具和用户体验保持一致,同时不隐藏运营风险或证据。
关键创新是把架构当成产品界面:能力通过责任、证明路径和交付边界来说明。
Bounded Data Agent 展示了产品架构、智能体系统、数据边界、控制机制和交付纪律上的应用工程判断。
Bounded Data Agent 是关于受治理的检索增强数据智能体平台的案例研究。受治理的检索增强数据智能体平台 其深度在于让边界、证据、工作流和交付选择变得可读。
Bounded Data Agent 通过清晰职责来组织,避免用户、工具、数据和运行时动作混成一次不透明的模型调用。
该流程将规划、工具使用、状态和结果呈现区分开,让操作者能理解发生了什么以及为什么发生。
页面把证明、日志、产物和评审路径视为产品的一部分,而不是可选的实现备注。
TypeScript 支撑实现叙事,周边技术栈承载持久化、界面、自动化和验证。
技术理由: Keeps the planner, tool executor, policy objects, and answer contracts typed end to end.
技术理由: Fits the HTTP and CLI runtime while staying close to the TypeScript agent framework.
技术理由: Acts as the governed warehouse, catalog store, document index, and audit-friendly source of truth.
技术理由: Keeps semantic retrieval inside Postgres so vector search and relational governance share one boundary.
技术理由: Provides typed schema, migrations, and seeds without hiding the SQL policy layer.
技术理由: Makes the data agent reproducible with Postgres and local services available for testing.
技术理由: Gives fast unit and evaluation checks for policy, query, and verifier behavior.
技术理由: Matches the explicit retrieve, sample, execute, verify, answer pipeline shape.
已实现工具: Asks the business semantic layer before touching raw warehouse tables.
已实现工具: Combines keyword and vector retrieval across documents, graph data, and embeddings.
已实现工具: Chooses representative data before the agent attempts broader analysis.
已实现工具: Inspects distributions and data quality so answers can state limits honestly.
已实现工具: Allows only guard-approved, scoped, read-only SQL to execute.
已实现工具: Checks results deterministically before the answer writer responds.
已实现工具: Scores fixture runs against thresholds so quality can be tracked.
已实现工具: Records the question, plan, tool calls, SQL, and evidence for review.