Hackeroom — give Claude Code a dev team you hire yourself, and watch them work in an office

Abhinav — Sun, 23 Aug 2026 07:37:45 +0000

Dev House

I got tired of telling one Claude Code session to plan, code, review, test, and fix its own mistakes.

So I built a software team around Claude Code.

You literally hire the team.

Give each member a name, role, model, permissions, and skills — then put them into seats like:

Planner → Coder → Reviewer → Tester → Auditor

Each member runs in its own real Claude Code session.

And you can watch the whole thing happen in a pixel-art office.

The weird part is that the team isn't hardcoded.

Want 3 coders and no reviewer? Do it.

Want an auditor that can only inspect? Do it.

Leave a seat empty? That phase simply doesn't run.

A few things I went surprisingly deep on:

Permissions actually enforce themselves. A PreToolUse hook checks every tool call against a capability manifest. .claude/ is unwritable, so an agent can't modify its own permissions.
Skills are isolated per agent. A huge skill document doesn't get dumped into everyone's context. It's loaded only when that member needs it.
Memory is shared without sharing conversations. Agents write down what they learn, and other agents get a tiny indexed summary before reading the full memory on demand.
Token usage is tracked per member. You can see exactly who's burning tokens and which model they're using.

So instead of:

You → Claude → everything

it's more like:

You → Team → Planner → Coder → Reviewer → Tester

And yes, there are caveats.

The part I'd actually love people to tear apart is the permission/hook architecture. I'd especially like feedback on whether the capability-manifest + PreToolUse approach is actually robust enough, and where it could be bypassed or improved.

Repo: https://github.com/SinghAbhinav04/dev-house

If you find it interesting, give the repo a star. It'd mean a lot.

I Built Persistent Memory for Claude Code Because My AI Kept Forgetting My Codebase

Abhinav — Sat, 08 Aug 2026 17:53:00 +0000

I use Claude Code heavily, and one thing kept annoying me: every new session felt like starting over.

I'd spend an entire session explaining my codebase — why a weird function exists, how the auth flow works, which API decision was intentional, and which "obvious" change would actually break everything.

Then I'd start a new session and Claude had forgotten all of it.

So I built OmniMemory: a local, Git-aware memory layer for coding agents.

The idea is simple: instead of every session starting with "here's my codebase, figure it out", the agent can start with "here's what we've already learned about this codebase."

OmniMemory stores things like architectural decisions, project context, workflows, gotchas, and other useful information from previous sessions.

The Git integration is the part I care about most. Memory is associated with branches and Git history, so context discovered while working on one branch doesn't have to become permanent global knowledge.

There's also a freshness problem with persistent memory. If the agent remembers that a function behaves a certain way, and I completely rewrite that function three weeks later, that memory shouldn't automatically remain trusted.

OmniMemory can check memories against changes in the codebase and identify potentially stale context. It can also use tree-sitter for symbol-level code analysis instead of treating an entire file as changed whenever one thing is modified.

I also wanted this to stay local. The memory store uses SQLite and doesn't require sending project memory to an external service.

For retrieval, OmniMemory uses BM25F-based ranking to find relevant memories instead of dumping the entire memory history into every prompt.

It integrates with Claude Code through hooks, so memory can be captured and injected into the normal coding workflow. There's also a local UI where you can inspect what the system actually remembers.

I'm building this around a simple idea: coding agents shouldn't just understand our codebase for one session. They should be able to accumulate useful knowledge about it over time.

If you use Claude Code or other coding agents heavily, I'd love to hear how you're currently dealing with the "AI forgot everything from yesterday" problem.

GitHub: https://github.com/SinghAbhinav04/Omni-Memory

DEV Community: Abhinav

Hackeroom — give Claude Code a dev team you hire yourself, and watch them work in an office

Dev House

I Built Persistent Memory for Claude Code Because My AI Kept Forgetting My Codebase