sec-guidance

Extract management guidance and forward-looking statements from SEC 10-K/10-Q filings using a local RAG pipeline.

SEC Guidance Extractor

Extracts forward-looking management guidance from SEC 10-K/10-Q filings using a local multi-modal RAG pipeline (Elasticsearch + sentence-transformers + Ollama).

The pipeline directory is read from the SEC_PIPELINE_DIR environment variable.

When to invoke

• User asks what management said about future revenue, earnings, margins, or outlook
• User asks for guidance from a 10-K or 10-Q filing
• User asks "what is [company]'s guidance for next quarter?"
• User wants forward-looking statements or risk factors from SEC filings

Setup (first time only)

1. Set the env var

In ~/.openclaw/openclaw.json:

json5

{
  skills: {
    entries: {
      "sec-guidance": {
        enabled: true,
        env: { SEC_PIPELINE_DIR: "/path/to/your/sec-rag-pipeline" }
      }
    }
  }
}

Or export in your shell before starting the gateway:

bash

export SEC_PIPELINE_DIR="/path/to/your/sec-rag-pipeline"

2. Start Elasticsearch

bash

docker ps | grep elastic   # confirm it's running

3. Download and index filings

bash

cd "$SEC_PIPELINE_DIR"

# Download filings (any ticker, any form type)
python download_sec_filings.py --ticker AAPL --form 10-Q --format pdf
python download_sec_filings.py --ticker AAPL --form 10-K --format pdf

# Ingest into Elasticsearch
python run_pipeline.py ingest

Check index status:

bash

python run_pipeline.py stats

Run guidance extraction

Default (5 standard guidance queries):

bash

python "{baseDir}/extract_guidance.py"

With options:

bash

python "{baseDir}/extract_guidance.py" --ticker AAPL --form 10-K --top-k 8

Single custom question:

bash

python "{baseDir}/extract_guidance.py" --query "What did management say about iPhone revenue next year?"

Or use the pipeline's ask command directly:

bash

python "$SEC_PIPELINE_DIR/run_pipeline.py" ask "What is management's outlook for gross margins?"

Add a new ticker

bash

cd "$SEC_PIPELINE_DIR"
python download_sec_filings.py --ticker MSFT --form 10-Q --format pdf --limit 8
python run_pipeline.py ingest

Standard guidance queries the script runs

1. Future revenue and earnings outlook
2. Forward-looking performance expectations
3. Expected gross margins and profitability
4. New product launches, services, or expansion plans
5. Macroeconomic risks and uncertainties for upcoming quarters

Output format

Each query prints the LLM answer with inline citations [1], [2], etc., plus a source list (filename, page number, filing date).

Pipeline requirements

Component	Purpose
Elasticsearch (local/Docker)	Hybrid search index
sentence-transformers	Embedding model (all-MiniLM-L6-v2)
cross-encoder	Neural reranker
Ollama (hermes3:8b or similar)	Answer generation
pymupdf, pdfplumber	PDF ingestion