npm install -g @kaelio/ktx
Then run ktx setup to start the interactive configuration wizard.
然后运行 ktx setup 启动交互式配置向导。
Scan PostgreSQL, Snowflake, BigQuery, ClickHouse, MySQL, SQL Server, SQLite — extract schemas, foreign keys, profiles.扫描 PostgreSQL、Snowflake、BigQuery、ClickHouse、MySQL、SQL Server、SQLite — 提取 schema、外键、数据画像。
YAML-based measures, dimensions, joins & segments. Validate, search, and compile SQL from approved definitions.基于 YAML 的度量、维度、连接和分段。验证、搜索并从已批准的定义编译 SQL。
Structured markdown pages with frontmatter (tags, refs, usage mode). Full CRUD + hybrid search.带 frontmatter（标签、引用、使用模式）的结构化 Markdown 页面。完整 CRUD + 混合搜索。
Reciprocal Rank Fusion across semantic, lexical, dictionary & token lanes. Configurable weights, k=60 default.跨语义、词法、字典和 token 通道的倒数排名融合。可配置权重，默认 k=60。
Model Context Protocol tools for agents: search, read, query, execute SQL (read-only enforced).为智能体提供模型上下文协议工具：搜索、读取、查询、执行 SQL（强制只读）。
Analyze historic SQL from Snowflake, BigQuery, PostgreSQL — detect table usage, common patterns, scope membership.分析来自 Snowflake、BigQuery、PostgreSQL 的历史 SQL — 检测表使用、常见模式、范围成员。
Multi-signal join detection: name similarity, type compatibility, value overlap, embedding similarity, profile uniqueness.多信号连接检测：名称相似度、类型兼容性、值重叠、嵌入相似度、画像唯一性。
Fetch → chunk → work units → reconcile → finalize. Isolated git worktrees, collision-safe with timestamp suffixes.获取 → 分块 → 工作单元 → 协调 → 完成。隔离的 git worktree，带时间戳后缀防冲突。
# Initialize configuration
ktx setup
# Ingest a PostgreSQL database
ktx ingest postgresql://user:pass@host:5432/mydb
# Search the semantic layer
ktx search "monthly active users by region"
# Start MCP server for agent integration
ktx mcp --port 3100
# List all semantic layer sources
ktx semantic listscore = Σ (weightlane / (k + ranklane)) 得分 = Σ (权重通道 / (k + 排名通道))
Enter ranks for each lane (1-based). Click calculate to see the fused RRF score. 输入每个通道的排名（从1开始）。点击计算查看融合后的 RRF 得分。
Share
Submit to AI Directories