Why I Spent a Year Building Sentinel-IR?

Thanks for support..
Release: "Sentinel-IR-MCP v0.2.2 on GitLab" (https://gitlab.com/jakubrojicek11/sentinel-ir-mcp/-/tree/v0.2.2?ref_type=tags)

Package: "sentinel-ir-mcp@0.2.2 on npm" (https://www.npmjs.com/package/sentinel-ir-mcp/v/0.2.2)

Imagine hiring a new assistant every morning.

Every day, you must explain your project again. What it does, why decisions were made, where the problems are, and what happened yesterday.

Now imagine hiring ten assistants. All ten need to rediscover the same things.

That is part of the problem I want to solve.

AI agents can write code, investigate bugs, and review projects. But giving them more tools doesn't automatically give them better context. And repeatedly feeding them entire codebases can waste time, tokens, and money.

That's why I built Sentinel-IR.

Think of a library with thousands of books. Instead of making every visitor read every book, the librarian provides a compact index of what's inside and where to look next.

Sentinel-IR does something similar for JavaScript code. It extracts structured facts about a codebase so an AI agent can get an initial map before deciding what deserves deeper investigation.

It doesn't replace the original code. It helps the agent find its way through it.

I have spent roughly a year exploring a bigger idea: what would AI development look like if agents could share useful, verifiable knowledge instead of repeatedly rediscovering the same things?

I don't claim to have solved that problem.

I want to make it cheaper to investigate, easier to verify, and possible to build on what has already been learned.

Because the future of AI may depend not only on how intelligent our agents become, but on how much useful work they can build on together.

That's why Sentinel-IR exists.

And I'm still exploring how far this idea can go.