Invite-only beta

Claude Code is brilliant,
but it forgets.

Every decision you made, every dead end you ruled out, the knowledge you earned getting there. context-memory keeps all of it, and brings it back the moment it’s relevant.

We’re inviting a first cohort by hand.

No spam free during the beta

For individual developers today. Team features coming soon.

↳ recalled automatically The work you’d otherwise re-explain.

Decision Dates are stored UTC, formatted at the edge. Don’t persist local time.

Dead end Cached the live order status, but most entries got invalidated before they were ever read. Read straight from Postgres.

Gotcha Migrations must run before deploy. The app boots but 500s if the schema is behind.

How it works

It runs while you code. No new habits, no file to maintain.

01Capture

As you and Claude Code work, it automatically records the decisions, dead ends, and gotchas you settle.

02Rank

Each memory is scored by what’s proven useful, so the noise sinks and the load-bearing memories rise.

03Recall

When something you’ve settled becomes relevant again, the right memories surface automatically. On that turn, not just at session start.

What context-memory solves

Every problem below is just a memory your agent didn’t keep.

No more re-explaining your stack and conventions in every new thread.
No more re-trying approaches you’ve already ruled out.
No more watching Tuesday’s hard-won fix vanish by Thursday.
No more cold starts that re-learn the same quirks.
No more CLAUDE.md files rotting as the code moves on.
No more RAG that hands back documents instead of the decisions you’ve made.

The difference

Same task, two sessions. One remembers what you ruled out last week.

Without context-memory Cold start

YouThe order status page feels slow. Can you speed it up?

Claude CodeI’ll add a Redis cache in front of the status query.

YouWe tried that last week. It made reads slower, since most entries got invalidated before they were read.

Claude CodeGot it. I’ll revert and index the query instead.

With context-memory Remembers

Dead endCached the live order status, but most entries were invalidated before read. Use Postgres.

YouThe order status page feels slow. Can you speed it up?

Claude CodeCaching was already ruled out here. It added latency, so I’ll index the status query instead.

Stop solving the same problem twice.

Request access →

FAQ

Product

How is this different from a CLAUDE.md file?

A CLAUDE.md is a file you maintain by hand, read in full every session, and that goes stale the moment the code moves on. context-memory captures knowledge automatically as you work, surfaces only what’s relevant to the task in front of you, and ranks it by what’s proven useful. Use both: CLAUDE.md for stable rules, context-memory for the knowledge that accumulates.

How is this different from Claude’s built-in memory?

Claude’s memory is about you: your preferences and how you like to work, carried across your conversations. context-memory is about your project. It holds the decisions, dead ends, and gotchas specific to a codebase, scoped per repo and captured automatically as you code. It’s a structured, inspectable store you can search, edit, and (soon) share with your team. The two are complementary: one remembers the developer, the other remembers the code.

What does it store? Does it read my whole codebase?

It stores the decisions, dead ends, and gotchas that surface as you work, not your source tree. Every memory is visible and editable: review, correct, or delete any of it.

Do I have to change how I work?

No. Install the plugin once; after that, capture and recall happen on their own. You keep using Claude Code exactly as you do now.

Compatibility

Does it work with tools other than Claude Code?

It’s built for Claude Code, where capture and recall happen automatically through its hooks. The memory itself is a standard MCP server, so other MCP-capable tools, including Claude Desktop, can read and write the same store. But the hands-off, automatic experience is Claude Code today.

Access

Who is context-memory for?

Individual developers who use Claude Code and are tired of their agent forgetting: solo devs, and anyone juggling several repos at once.

Is it private? Can my team use it?

Your memory is private to you. Team features are shipping soon: shared memory across a whole team’s agents, where you choose what to share. Join the waitlist and mention your team, and we’ll reach out when team access opens.

Is it free?

It’s free during the beta. Pricing comes later, and we’ll give notice before anything changes.