Your AI Agent Ordered Bananas. Here's Why.

Kriday Dave — Fri, 15 May 2026 10:41:27 +0000

Your AI Agent Ordered Bananas. Here's Why.
You told your AI agent you like apples and oranges. Somewhere between Agent 1 and Agent 2, a banana snuck in. Now your grocery agent is ordering bananas. Not because the LLM hallucinated -- because the context handoff was broken.
This is the real problem nobody talks about. Everyone's arguing about which model is smarter while the plumbing underneath is leaking.
We blame the LLM for everything
Yelling at Claude at 3AM to fix your broken pipeline isn't going to work. Claude doesn't know you. Claude doesn't remember what Agent 1 said to Agent 2 three steps ago. Each handoff between agents is just a JSON blob getting passed around with zero guarantees. No signing. No validation. No rollback if something goes wrong.
RAG gets thrown around as the solution to everything but RAG just helps agents find relevant information. It doesn't protect the information that's already been passed around and corrupted.
If your AI agent corrupts your context -- your Obsidian vault, your memory store, whatever you're using -- every downstream agent inherits the damage. Silently.
The banana problem, formally
Here's what actually happens without any protection:
python# Agent 1 produces output
agent1_output = {"entities": ["Apple", "2024 revenue"], "summary": "Apple grew"}

Manual serialization -- easy to lose data, corrupt context

context = json.dumps(agent1_output)

Agent 2 receives corrupted context

agent2_input = f"Given: {context}\nAnalyze this."
Agent 2 might drop keys. Add hallucinated ones. Quietly rewrite history. You won't know until something downstream breaks in a way that's nearly impossible to trace back.
What if we treated context handoffs like database transactions?
That's the question that led to Relay.
Relay treats your context like an append-only ledger. Every handoff between agents gets signed and validated. If an agent drops a critical key -- like "entities" disappearing between steps -- Relay catches it and rolls back to the last clean state automatically.
pythonfrom relay.core_pipeline import CoreRelayPipeline

pipeline = CoreRelayPipeline(
signing_secret="your-secret-key",
token_budget=8000
)

Agent 1 creates a signed envelope

result = pipeline.execute_step({"entities": ["Apple"], "revenue": "2024"})
envelope1 = result.value # signed, immutable

Agent 2 accidentally drops "entities"

result = pipeline.execute_step({"summary": "growth"}) # contradiction detected

Relay rolls back automatically

result = pipeline.rollback()
restored_envelope = result.value

Back to the clean state from step 1

No banana.
What Relay actually does
Every context move between agents gets wrapped in a signed envelope with a version, a timestamp, the token budget, and an HMAC signature. If anything changes, the signature fails. If a critical key disappears, the validator fires. If the validator fires, you get a rollback to the last clean checkpoint instead of silent corruption propagating through your pipeline.
It also supports parallel agents through fork-join -- you can run multiple agents against the same context snapshot at the same time, then merge or vote on their outputs. If there's a conflict, the whole step rolls back cleanly.
Why this matters
Reliable AI systems need reliable infrastructure underneath them. The model quality is only as good as the context it receives. Right now most multi-agent pipelines have no protection on context handoffs at all.
I built Relay as a 10th grader because I kept running into this problem and couldn't find anything that solved it properly. Context matters. It needs to be protected, not just passed around and hoped for the best.
You can use it today:
pip install relay-middleware
Or check the source: https://github.com/kridaydave/Relay

I'm 15 and Built a Production-Grade MCP Server During Winter Break

Kriday Dave — Sun, 08 Feb 2026 15:31:08 +0000

The Problem: My Downloads Folder Was a Disaster 🗑️
I'm a 15-year-old student, and like most people, my Downloads folder was a mess.
1,200+ files scattered everywhere. Videos mixed with documents. Installers next
to screenshots. It was chaos.
Every time I needed to find something, I'd waste 10 minutes scrolling through
hundreds of randomly named files. There had to be a better way.

The Discovery: NetworkChuck's MCP Tutorial 🎥
During winter break, I was browsing YouTube and stumbled on NetworkChuck's video
about the Model Context Protocol (MCP). He showed how you can give AI assistants
like Claude new capabilities through custom servers.
That's when it clicked: What if Claude could organize my files for me?
I spent the next week building what would become File Organizer MCP.

The Build: From Idea to Production in 7 Days 🚀
Day 1-2: Learning & Planning

Studied the MCP specification
Learned about file categorization algorithms
Planned the architecture

Day 3-4: Core Features

Built file scanning and categorization
Implemented duplicate detection using SHA-256 hashing
Added dry-run mode for safety

Day 5-6: Polish & Security

Migrated everything to TypeScript
Added 8-layer path validation (security is important!)
Wrote comprehensive tests (ended up with 268!)
Created an interactive setup wizard

Day 7: Launch

Published to NPM
Listed on MCP Registry
Wrote documentation
Hit publish 🎉

The Tech Stack 🛠️
Why I chose these technologies:

TypeScript - Type safety prevents bugs
Zod - Runtime validation for user inputs
Jest - Comprehensive testing framework
Node.js - Works everywhere

Security was a priority:

8-layer path validation pipeline
TOCTOU mitigation using file descriptors
Atomic operations to prevent data loss
Sandboxed operations (files can't escape allowed directories)

The Features ✨
File Organizer MCP provides these tools to Claude (and other AI assistants):

Smart Categorization - Automatically sorts files into 12+ categories
Duplicate Detection - Finds duplicate files and calculates wasted space
Batch Renaming - Rename multiple files with patterns and regex
Metadata Extraction - Reads EXIF/ID3 tags for smart organization
Scheduled Organization - Cron-based auto-organization
Rollback Support - Undo operations if needed

All with a simple command:
bashnpm i file-organizer-mcp
npx file-organizer-mcp --setup
The setup wizard auto-detects your AI clients (Claude Desktop, Cursor, Windsurf,
etc.) and configures them for you. One command, done.

The Results: 728 Downloads in Week 1 📈
I honestly didn't expect much. This was a personal project to solve my own problem.
But within the first week:

✅ 728 npm downloads
✅ 11 GitHub stars
✅ Listed on MCP Registry
✅ Users from around the world

People were actually using it! And more importantly, it was helping them.

What I Learned 💡
Technical Lessons:

Security is hard - Path traversal, symlink attacks, race conditions...
there's so much to think about!
Testing matters - Those 268 tests caught SO many bugs
TypeScript is worth it - Caught errors at compile time instead of runtime
Documentation is for future you - I've already forgotten why I wrote some code

Personal Lessons:

Just ship it - Don't wait for perfect, ship and iterate
Solve real problems - Personal pain points make the best projects
Open source is rewarding - Seeing others use your code is amazing
Age doesn't matter - Good code is good code, regardless of who wrote it

What's Next? 🔮
Currently working on:

Cloud storage integration (S3, Dropbox, Google Drive)
More metadata support (organize photos by location, music by genre)
Better scheduling (more flexible automation)
Performance improvements (handle even larger directories)

Try It Yourself 🚀
Want to organize your files with AI?
bashnpm i file-organizer-mcp
npx file-organizer-mcp --setup
Then ask your AI assistant:

"Organize my Downloads folder"
"Find duplicate files in Documents"
"Show me my largest files"

Links:

📦 NPM: https://www.npmjs.com/package/file-organizer-mcp
🐙 GitHub: https://github.com/kridaydave/File-Organizer-MCP
📚 MCP Registry: https://registry.modelcontextprotocol.io/servers/io.github.kridaydave/file-organizer

Final Thoughts 💭
This started as a simple solution to a messy Downloads folder. It became a
production-grade tool with hundreds of users.
If you're reading this and thinking "I could never build something like this" -
you absolutely can. I'm just a 15-year-old who watched a YouTube video and had
a problem to solve.
Start small. Solve your own problems. Ship it. Iterate.
The hardest part is starting. So start today.

Questions? Comments?
Drop a comment below or open an issue on GitHub. I read everything!
And if you try File Organizer MCP, let me know what you think. Your feedback
helps make it better for everyone.
Happy organizing! 🎯
Tags to add:

mcp #claude #ai #typescript #opensource #nodejs #productivity #tools

P.S. - Yes, I really am 15. And yes, this really was built during winter break.
And yes, you can do this too. Now go build something cool! 🚀

Forem: Kriday Dave