Breaking

Open-Source AI Agent Scores Stocks From SEC Filings and Social Sentiment

July 1, 2026 at 19:26 EDT

AI Agents
Finance
Open Source

Developer Ivan Brigida has released stocks-scoring-agent, an open-source AI agent that jointly analyzes U.S. companies' SEC filings and social media sentiment. Built as a multi-agent system with 17 analysis tools, its code and usage examples are published in a GitHub repository.

Open Source · Apache-2.0

stocks-scoring-agent: one AI workflow fusing SEC filings with social sentiment

A Python research agent built on the OpenAI Agents SDK (gpt-5.4-mini) that bundles two jobs once done separately — mechanically parsing 10-K/10-Q documents and gauging investor mood on X and Reddit — into a single pipeline. For research and education only; not investment advice.

Total tools in the agent

Agent modes for different uses

JSON fields in structured output

How the 17 tools break down

Each block = one tool · column height is proportional to the count

Fundamentals

Search / screening

Social sentiment

SEC filings

The core pipeline

Two separate research jobs, merged into one output

SEC filings
EDGAR 10-K / 10-Q

Social sentiment
X · Reddit

→

Scored analysis
30-field JSON

Four agent modes

SimpleAgent

Stateless, single-shot analysis

ConversationAgent

With memory, automatic ticker tracking

StructuredAgent

Narrative plus 30-field JSON (Pydantic)

FreeAgent

Supports local Ollama models

What developers value

Cuts research time by merging document summarization and sentiment
Strict schema output is easy for downstream systems
Runs locally via Ollama if desired

What critics flag

Generative-model hallucinations / misinformation
Data latency in fetched sources
U.S.-centric market constraint

Continue reading

The rest of this article is for AI News Blitz readers. Choose an option below to keep reading.