DEV Community

Jamse Bao
Jamse Bao

Posted on

Tried `K-Dense-AI/scientific-agent-skills` Today: A Hands-on Review for AI Developers

#ai #opensource #programming #tutorial

Tried K-Dense-AI/scientific-agent-skills: An AI Scientist in Your Agent Stack

K-Dense-AI/scientific-agent-skills is gaining serious traction—up 498 GitHub stars today—and the reason is practical: it turns a general-purpose AI agent into a research-focused assistant without forcing you to build every workflow from scratch.

The project includes 163 validated skills and access to more than 100 scientific databases across biology, chemistry, medicine, and drug discovery. It is designed for scientists, but the developer experience is equally interesting: the library works with Cursor, Claude Code, Codex, Pi, Antigravity, and the open Agent Skills standard.

The useful distinction is that this is not just a prompt collection. It provides reusable capabilities for literature research, data analysis, molecular workflows, and scientific investigation—areas where generic agents often produce shallow or poorly structured results.

A quick OpenAI-compatible gateway setup could look like this:

export OPENAI_BASE_URL="https://b-lost.com/v1"
export OPENAI_API_KEY="$BLOST_API_KEY"

curl "$OPENAI_BASE_URL/chat/completions" \
  -H "Authorization: Bearer $OPENAI_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "claude-fable-5",
    "messages": [
      {
        "role": "user",
        "content": "Use the scientific skills to compare recent CRISPR delivery methods."
      }
    ]
  }'
Enter fullscreen mode Exit fullscreen mode

For long research sessions, Prompt Caching is especially relevant. B-Lost supports native Anthropic /v1/messages caching, with a 90% discount on cache hits. That can reduce repeated costs when the same system instructions, skill definitions, or reference context are sent across multiple agent steps.

The relay also offers 20% off official list pricing, but the bigger engineering win is keeping the endpoint configurable. You can test the same scientific workflow through Cursor, Cline, Roo Code, Windsurf, Aider, or LibreChat without rewriting your agent integration.

Repository: https://github.com/K-Dense-AI/scientific-agent-skills

Top comments (0)