# Prospective GSoC Contributor: Introduction & Agentic / MCP Tooling for AiiDA

**URL:** <https://aiida.discourse.group/t/prospective-gsoc-contributor-introduction-agentic-mcp-tooling-for-aiida/777>\
**Category:** General Usage\
**Created:** [November 2, 2026, 9:40am UTC](https://aiida.discourse.group/t/prospective-gsoc-contributor-introduction-agentic-mcp-tooling-for-aiida/777 "2026-11-02T09:40:21Z")\
**Posts on this page:** 2\
**Page:** 1

<div class="post-metadata">

**Author:** ![KarimmYasser](https://yyz2.discourse-cdn.com/free1/user_avatar/aiida.discourse.group/karimmyasser/32/468_2.png) [@KarimmYasser](https://aiida.discourse.group/u/KarimmYasser)\
**Post date:** [November 2, 2026, 9:40am UTC](https://aiida.discourse.group/t/prospective-gsoc-contributor-introduction-agentic-mcp-tooling-for-aiida/777/1 "2026-11-02T09:40:21Z")

</div>

Hi everyone! 👋

I’m Karim Yasser, a Computer Engineering student at Cairo University, and I’m looking forward to contributing to the AiiDA ecosystem for Google Summer of Code (under NumFOCUS).

I’ve been working on autonomous agent harnesses and scientific workflow automation. Recently, I engineered a 19-tool Model Context Protocol (MCP) server with deterministic C++ renderers and pre-flight validation gates for OpenFOAM CFD simulations (preventing syntax hallucinations and boundary errors before launching long HPC runs).

I set up `aiida-agents` locally, verified the test suite across both macOS Apple Silicon and Windows, and opened **PR #111** ([aiidateam/aiida-agents#111](https://github.com/aiidateam/aiida-agents/pull/111)) to configure `pythonpath` in `pyproject.toml` for seamless test discovery.

I am particularly excited about the **AiiDA Copilot / Agentic Workflow** direction. I’m currently studying how the MCP tools interact with AiiDA’s provenance graph and would love early feedback on two architectural design ideas for the project proposal:

1. **Pre-flight Validation & Bounded Dispatch:** Adding a lightweight simulation parameter guardrail layer (analogous to static dictionary checking) to catch unphysical or syntax-invalid inputs before jobs are committed to HPC clusters.
2. **Provenance-Aware Agent Auditing:** Attaching prompt metadata, model versions, and tool execution traces directly into calculation node extras for complete provenance reproducibility.

Looking forward to discussing with the maintainers and collaborating with the community!

Best regards,  
Karim Yasser  
[GitHub](https://github.com/KarimmYasser) | [Portfolio](https://karim-yasser.vercel.app)

---

<div class="post-metadata">

**Author:** ![supermaterials](https://avatars.discourse-cdn.com/v4/letter/s/ec9cab/32.png) [@supermaterials](https://aiida.discourse.group/u/supermaterials)\
**Post date:** [November 3, 2026, 6:29am UTC](https://aiida.discourse.group/t/prospective-gsoc-contributor-introduction-agentic-mcp-tooling-for-aiida/777/2 "2026-11-03T06:29:26Z")

</div>

This post was flagged by the community and is temporarily hidden.