Retrieve information from simulator manual and example DATA files. Use when answering keyword format questions, syntax queries, or when looking up official documentation and working examples. Essential for understanding keyword definitions, parameter tables, and concrete usage patterns.
This skill provides retrieval tools for accessing simulator documentation and example files through vector search.
The RAG skill enables agents to:
Retrieves information from the simulator manual and official documentation using semantic search.
Usage:
simulator_manual(query: str) -> str
Parameters:
query: Natural language query about keywords, syntax, or documentation (e.g., "COMPDAT keyword format", "WELSPECS syntax and fields")Example:
simulator_manual("What is the COMPDAT keyword format?")
Returns: Retrieved documentation snippets from the simulator manual with source citations.
When to use:
parse_simulation_input_file (Section 2.4) to get keyword contextRetrieves example DATA files and case studies using semantic search.
Usage:
simulator_examples(query: str) -> str
Parameters:
query: Natural language query about keyword examples or usage patterns (e.g., "COMPDAT keyword format", "WCONINJE injection rate examples")Example:
simulator_examples("COMPDAT keyword format")
Returns: Retrieved example DATA file snippets showing concrete keyword usage.
When to use:
simulator_manual when manual lacks format details or examples (Section 2.5)simulator_manual to get example context for modifications (Section 2.4)This skill integrates with the Simulator Agent's decision tree (TOOL_DECISION_TREE.md):
Keyword Q&A Flow (Section 2.5):
simulator_manual → (simulator_examples) → answer
Scenario Test Chain (Section 2.4):
parse_simulation_input_file → simulator_manual (inferred keyword) → simulator_examples (same keyword) → modify_simulation_input_file → run_and_heal
Keyword Chain (Section 3.4):
simulator_manual → simulator_examples → synthesize format + example → final answer
Tools are implemented as LangChain retriever tools with:
scripts/extract_keyword.py provides RAG + LLM keyword extraction (e.g. "plot field oil" → FOPT). Used by plot_skill validators and other skills that need to infer keywords from natural language.
from simulator_agent.skills.rag_skill.scripts.extract_keyword import extract_keyword
kw = extract_keyword("plot field cumulative oil production", intent="summary_metric") # -> "FOPT"
RAG tools use Milvus collections created by ./scripts/setup.sh --full:
| Tool name | Milvus collection | Ingested by |
|---|---|---|
simulator_manual | docs | ingest_papers.sh |
simulator_examples | simulator_input_examples | ingest_opm_examples.py |
Environment variables:
MILVUS_URI: Milvus endpoint (default: http://localhost:19530; Docker: http://standalone:19530)NVIDIA_API_KEY: Required for embeddings, reranker, and LLM