agapi 2025.12.25

agapi

🌐 AtomGPT.org API (AGAPI)

AGAPI provides a simple way to interact with AtomGPT.org, enabling Agentic AI materials science research through intuitive APIs.

A significant amount of time in computational materials design is often spent on software installation and setup — a major barrier for newcomers.

AGAPI removes this hurdle by offering APIs for prediction, analysis, and exploration directly through natural language or Python interfaces, lowering entry barriers and accelerating research.

---

📖 Table of Contents

• API Docs
• 🧠 Capabilities & Example Prompts
  • 1️⃣ Access Materials Databases
  • 2️⃣ Graph Neural Network Property Prediction
  • 3️⃣ Graph Neural Network Force Field
  • 4️⃣ X-ray Diffraction → Atomic Structure
  • 5️⃣ Live arXiv Search
  • 6️⃣ Web Search
  • 7️⃣ Visualize Atomic Structures
  • 8️⃣ General Question Answering
  • 9️⃣ Structure Manipulation
  • 🔟 Voice Chat Interaction
• 🚀 Quickstart
  • Colab Notebook
  • Python SDK
• 🎥 YouTube Demos
• 📚 References
• ❤️ Note

---

API Docs

Replace sk-XYZ with your API key from atomgpt.org>>account>>settings.

AtomGPT.org/docs

🧠 Capabilities & Example Prompts

AGAPI supports natural language interaction for a wide range of materials science tasks.
Each section below includes a prompt example and expected output.

---

1️⃣ Access Materials Databases

Prompt:

List materials with Ga and As in JARVIS-DFT

Response:
Displays all GaAs-containing entries from the JARVIS-DFT database.

---

2️⃣ Graph Neural Network Property Prediction (ALIGNN)

Prompt:

Predict properties of this POSCAR using ALIGNN

(Upload a POSCAR, e.g. example POSCAR file)

Response:
Returns AI-predicted material properties (formation energy, bandgap, etc.).

---

3️⃣ Graph Neural Network Force Field (ALIGNN-FF)

Prompt:

Optimize structure from uploaded POSCAR file using ALIGNN-FF

(Upload a POSCAR, e.g. example file)

Response:
Generates optimized structure and energy data.

---

4️⃣ X-ray Diffraction → Atomic Structure

Prompt:

Convert XRD pattern to POSCAR

(Upload an XRD file, e.g. example XRD file)

Response:
Predicts atomic structure that best matches the uploaded diffraction pattern.

---

5️⃣ Live arXiv Search

Prompt:

Find papers on MgB₂ in arXiv. State how many results you found and show top 10 recent papers.

Response:
Summarizes and lists the latest publications from arXiv related to MgB₂.

---

6️⃣ Web Search

Prompt:

Search for recent advances in 2D ferroelectric materials.

Response:
Fetches and summarizes up-to-date information from web sources on the requested topic.

---

7️⃣ Visualize Atomic Structures

Prompt:

Visualize the crystal structure of Silicon in 3D.

Response:
Generates a 3D interactive visualization of the given structure (CIF or POSCAR).

---

8️⃣ General Question Answering

Prompt:

Explain the difference between DFT and DFTB.

Response:
Provides a concise explanation with context and examples.

---

9️⃣ Structure Manipulation

Prompt:

Replace oxygen atoms with sulfur in this POSCAR.

Response:
Outputs a modified POSCAR file with requested atomic substitutions.

---

🔟 Voice Chat Interaction

Prompt (spoken):

What is the bandgap of silicon?

Response (spoken):

The bandgap of silicon is approximately 1.1 eV.

Enables voice-based chat for hands-free interaction with materials science tools.

The table below lists available endpoints, the corresponding module, and description.

Endpoint	Module / Function	Description
/materials/property	ALIGNN	Predicts materials properties such as formation energy, bandgap, and elastic moduli directly from structure files.
/materials/forcefield	ALIGNN-FF	Computes energies, forces, and stresses for structure relaxation and molecular dynamics simulations with near-DFT accuracy.
/materials/xrd	XRDStructurePrediction	Determines atomic structures from uploaded XRD files to identify crystal structures.
/literature/search	arXivSearchAgent	Retrieves and summarizes recent arXiv or web publications on specified research topics.
/visualization/structure	StructureViewer	Generates interactive 3D visualizations of input structures and enables atomic structure editing.
/database/jarvis	JarvisAPI	Provides direct access to JARVIS materials data and pre-computed properties for workflow integration.
/interface/voice	VoiceChat	Enables voice-based chat for hands-free interaction with AGAPI.
/literature/search	Crossref	Accesses publication metadata and citation information through the Crossref API.

---

🚀 Quickstart

Colab Notebook

Try AGAPI instantly in Google Colab:
👉 AGAPI Example Notebook

Python SDK

For detailed SDK usage:
👉 agapi/README.md

---

🎥 YouTube Demos

Watch AGAPI in action on YouTube:
🎬 AGAPI Demo Playlist

---

📚 References

1. AGAPI-Agents: An Open-Access Agentic AI Platform for Accelerated Materials Design on AtomGPT.org
2. ChatGPT Material Explorer: Design and Implementation of a Custom GPT Assistant for Materials Science Applications
3. The JARVIS infrastructure is all you need for materials design
4. AtomGPT: Atomistic Generative Pretrained Transformer for Forward and Inverse Materials Design

Full publication list

---

❤️ Note

“AGAPI (ἀγάπη)” is a Greek word meaning unconditional love.

DISCLAIMER

AtomGPT.org can make mistakes. Please verify important information.

We hope this API fosters open, collaborative, and accelerated discovery in materials science.