agentic-stack (.agent/ folder) vs context-mode
Both in the memory & context category. Side-by-side — pick the one that fits your stack tonight.
One portable folder of memory + skills that works across every AI agent tool.
- rating
- 3★
- tested
- —
- cost
- free
- install
- drop-in
- stars
- 1,357
- updated
- 5d ago
You only use one agent tool and never plan to switch.
Stashes noisy tool output off to the side. Claude only sees a clean summary. 98% lighter.
- rating
- 5★
- tested
- —
- cost
- free
- install
- drop-in
- stars
- 8,815
- updated
- 4d ago
You only run short, quiet sessions where context bloat isn't a problem.
why it matters · agentic-stack (.agent/ folder)
Your agent's memory, skills, and workflows usually get locked to whichever tool you started with — Claude Code, Cursor, Windsurf, whatever. Switch tools and you start over from zero. agentic-stack is a simple `.agent/` folder structure that holds everything — memory, skill files, protocols — in one place. Drop it in any project and every major agent tool reads from it. Swap Claude Code for Cursor next month and your lessons come with you.
why it matters · context-mode
When Claude Code runs something loud — a Playwright browser session, a GitHub API dump, a big test suite — the full output balloons your context window. Context Mode intercepts that, stashes the whole thing in a local database, and hands Claude just the clean summary. Your session gets 98% lighter without losing any info you might need later. You can still pull the full output by name when you want it. Works with Claude Code + 11 other platforms. Free.