AI coding agents forget everything between sessions. Scritty is a new tool pitching itself as the fix — a shared, searchable memory layer that sits across every AI coding agent you use.
The core idea:
- Persistent memory — context from past sessions carries forward instead of vanishing when a conversation ends
- Shared across agents — one memory store works whether you’re using Cursor, Claude Code, Copilot, or another agent
- Searchable — agents can query prior decisions, patterns, and project context instead of re-discovering them each run
The problem Scritty targets is real. Anyone running AI coding agents on a non-trivial codebase hits the same wall: the agent doesn’t remember that you prefer Tailwind over CSS modules, that the auth layer uses a custom middleware, or that the team decided last week to deprecate a particular API endpoint. You re-explain these things constantly. Multiply that across multiple agents and the friction compounds.
A shared memory layer could cut that repetition. It also opens a path toward agents that build up genuine project understanding over time — something closer to a junior dev who’s been on the team for months rather than one who showed up this morning.
NOTE
Scritty launched on Product Hunt with minimal public documentation. Pricing, supported agents, and integration details aren’t fully clear yet. Treat this as an early-stage tool worth watching, not one to build workflows around today.
The competitive space here is thin but growing. Cursor has its own .cursor/rules files for persistent context. Claude Code supports CLAUDE.md project files. But these are agent-specific and not searchable in the way Scritty describes. A cross-agent memory layer is a different bet — it assumes developers will keep using multiple agents rather than consolidating into one.
Scritty is live on Product Hunt. If you’re running two or more AI coding agents on the same project, this is the category to track.