菜单
Use when the user wants to build a RAG (retrieval-augmented generation) pipeline with IRIS as the vector store
用 Codex 或 Claude 帮你安装 复制这段 Prompt，粘贴到 Codex、Claude 或其他助手里，让它检查 Skill 页面并帮你完成安装。
基于 SOC 职业分类
Use when creating or modifying InterSystems IRIS interoperability production components in Python — Business Services, Business Processes, Business Operations, Adapters, Messages, or Production definitions.
Measure ObjectScript line coverage using %Monitor.System.LineByLine. Wraps a %UnitTest.Manager.RunTest() call, collects per-class hit/total counts, prints a coverage table. Prerequisite: iris-coverage-setup (bbsiz=4096, IRIS restarted).
Measure ObjectScript line coverage using iris_coverage. Use when a user asks about test coverage, coverage reports, or wants to know how well their test suite exercises their code.
InterSystems AI Hub EAP (Early Access Program) — accurate API patterns for builds 158/159/161/162 (current). Covers %AI.Agent declarative Parameters, %AI.Provider.Create, ConfigStore/GetProviderForConfig, @{env/config/wallet} substitution, session management, streaming, tool sets, and known breaking changes from build 141. Load when helping EAP participants set up, build, or debug AI Hub projects.
Use after writing or editing any ObjectScript .cls file, after applying a fix to a compile error, or before declaring a task done — uploads the class to IRIS via Atelier REST, compiles it, and returns structured errors for immediate fixing.
Manage and observe IRIS Interoperability productions — lifecycle, logs, queues, and message tracing
| name | iris-vector-rag |
| description | Use when the user wants to build a RAG (retrieval-augmented generation) pipeline with IRIS as the vector store |
| managed_by | iris-agentic-dev |
Project: iris-vector-rag
Install: zpm "install iris-vector-rag"
Requires: IRIS 2024.1+; ZPM; an OpenAI/embedding provider API key
iris-vector-rag is a complete retrieval-augmented generation (RAG) framework built on IRIS:
Builds on iris-vector-graph for the vector storage layer; adds the full document pipeline on top.
User query
│
▼
Embed query (OpenAI / local model)
│
▼
IRIS vector search ──→ Top-K chunks (with metadata)
│
▼
Prompt assembly: [system] + [retrieved context] + [user question]
│
▼
LLM (OpenAI GPT-4o / Azure / local)
│
▼
Answer + citations
// Ingest a document
Set rag = ##class(community.vectorrag.Pipeline).%New()
Do rag.IngestFile("/data/mydoc.pdf", "my-collection")
// Query
Set answer = rag.Query("What does section 3 say?", "my-collection")
Write answer
zpm "install iris-vector-rag"
Set environment variables (or use IRIS credentials manager):
OPENAI_API_KEY=sk-...
IRIS_NAMESPACE=USER