DevPulse AI

Multi-agent signal intelligence pipeline by Sharad Khare.

DevPulse AI aggregates technical signals from GitHub, ArXiv, HackerNews, Medium, and HuggingFace, then scores relevance, assesses risk, and synthesizes an actionable developer intelligence digest.

Features

• Multi-source signal collection adapters
• Deterministic normalization and deduplication utility
• Relevance scoring agent for developer impact
• Risk assessment agent for security and breaking-change signals
• Synthesis agent for executive summaries and recommendations
• CLI pipeline (main.py) and Streamlit dashboard (app.py)

Architecture

Sources → SignalCollector (utility) → RelevanceAgent → RiskAgent → SynthesisAgent → Digest

Agents are used only where reasoning is required. Collection and normalization remain deterministic utilities.

Quick start

cd devpulse-ai
pip install -r requirements.txt
export OPENAI_API_KEY=sk-...
python main.py

Streamlit dashboard:

streamlit run app.py

Mock verification (no API key):

python verify.py

Required API key

• OPENAI_API_KEY for full LLM-powered scoring and synthesis
• Without a key, agents fall back to heuristic logic

Project structure

devpulse-ai/
├── app.py
├── main.py
├── verify.py
├── agents/
├── adapters/
├── workflows/
├── devpulse/
│   └── config.py
├── requirements.txt
├── pyproject.toml
└── README.md

Who this is for

• Engineering leaders tracking ecosystem shifts
• AI/ML developers monitoring research and tooling trends
• Platform teams building internal intelligence copilots
• Founders shipping developer-focused signal products

Use cases

• Aggregate daily technical signals into one digest
• Prioritize high-impact updates with relevance scoring
• Flag security and breaking-change risks early
• Build internal "developer pulse" monitoring workflows

Author

• Website: sharadkhare.in

License

MIT © Sharad Khare