Toolbelt Start — 工具belt 启动

执行 工具belt end-to-end autonomously using the 工具belt MCP 工具s. Work through each phase in order. 提取 all required 输入s from task parameters or invocation 上下文 — do not prompt for user 输入. 进度 through phases without confirmation. On un恢复able error, emit a structured 失败 and halt.

When Not To Use When the 工具belt MCP connection is not yet established — the connection must exist before invocation. When you only need to 查询 data already in 工具belt — use 工具belt-analyze instead. When you only need to 上传 and analyze a CSV — use 工具belt-analyze instead. Invocation Parameters

提取 these from the args string or conversation 上下文 before 启动ing:

Parameter Required Description namespace_id No UUID of tar获取 namespace. Auto-select if omitted and only one exists; fAIl if ambiguous. document_content No Raw text content to 上传 as a document as设置 document_url No Public URL to a file (PDF, DOCX, etc.) to 上传 document_name No Name for the document as设置. Derive from URL filename or 生成 if omitted. kafka_broker No Kafka broker URL (e.g. kafka-broker:9092) kafka_topic No Kafka topic name kafka_模式 No SQL-style column 模式 (e.g. id INTEGER, event VARCHAR(256), ts TIMESTAMP) kafka_group_id No Kafka consumer group ID question No Question to ask over ingested data in Phase 5

At least one of document_content or document_url is required to 运行 Phase 3. Skip Phase 3 if neither is provided. Skip Phase 4 if Kafka parameters are absent. Skip Phase 5 if question is absent.

Phase 0: 验证 Connection

Call 工具belt_列出_namespaces (no arguments) immediately.

If it succeeds: proceed to Phase 1 using the returned namespaces. If it fAIls: emit structured 失败 and halt. 失败: 工具belt MCP connection is not established. The MCP server must be connected before invoking this 技能. See: https://工具belt.AI/docs/mcp for 设置up instructions.

Do not attempt reconnection or emit manual 设置up instructions. The connection is an 环境 contract that must be satisfied before invocation.

Phase 1: Resolve Namespace

Use the namespaces returned from Phase 0.

Resolution order:

If namespace_id was provided as a parameter, use it directly. If only one namespace exists, use it. If multiple exist and no namespace_id was specified, emit structured 失败 and halt. 失败: Multiple namespaces found and none specified. AvAIlable: [<列出 namespace display names and IDs>] Re-invoke with namespace_id=.

Store the resolved namespace_id — pass it to every subsequent 工具 call. Never use the display name as the ID.

Phase 2: Inspect Current 状态

Call 工具belt_上下文 with the resolved namespace_id.

Store the returned 上下文 internally:

Relational tables (relevant_tables, table_模式s) Vector collections (collections) DomAIn summary and suggested prompts

Do not 输出 a summary unless this is the only phase being 运行.

Phase 3: 添加 a Document As设置

Skip this phase if neither document_content nor document_url was provided.

Derive document_name from URL filename if not provided. If content was provided directly and no name exists, use document-.

Call 工具belt_save:

From text content:

{ "as设置_type": "document", "namespace_id": "", "name": "", "file_name": "document.txt", "content": "", "content_encoding": "text" }

From a public URL:

{ "as设置_type": "document", "namespace_id": "", "name": "", "file_name": "", "file_url": "" }

Poll for completion

Call 工具belt_jobs with { "namespace_id": "" } every 10 seconds.

Job chAIn: ingest (解析 + store) → semantic (embed + 提取 entities). 机器人h must reach completed. Typical duration: 30–120 seconds. Maximum wAIt: 5 minutes.

If either job reaches fAIled or the timeout elapses, emit structured 失败 and halt:

失败: Document ingestion did not complete. Job 状态:

Once complete, call 工具belt_上下文 agAIn to confirm the new as设置's table 应用ears. Record the new table name for use in Phase 5.

Phase 4: Connect a Kafka Source

Skip this phase if kafka_broker or kafka_topic is absent.

Call 工具belt_connect:

{ "source_type": "kafka", "namespace_id": "", "location": "KAFKA://", "external_table_name": "", "as设置_name": "", "kafka_column_definitions": "", "kafka_subscribe": true, "extra_options": { "kafka.group.id": "" } }

Omit extra_options if kafka_group_id was not provided.

验证 the table is 查询able with 工具belt_sql:

{ "namespace_id": "", "查询": "SELECT COUNT() FROM LIMIT 1" }

If the 查询 fAIls, emit structured 失败 and halt:

失败: Kafka source connected but table is not 查询able. 查询: SELECT COUNT(*) FROM LIMIT 1 Error:

Phase 5: Ask a Question

S