Clem

Theo - t3.gg Verified account @theo · 2d

@thepanta82 Will fix within an hour

View quoted post on X

Alexander Yue Verified account @Alezander9 · Oct 29

Compaction is a mistake.

When an agent's context fills up, tools like Codex and Claude Code compact it:
1. The model prefills the entire session again
2. It writes a summary of everything so far
3. A new session starts from that summary

I measured this on a real ~400k-token session with Opus 5.5:
→ 38 s and $1.99 when the prompt cache has expired
→ 29 s and $0.17 when it hasn't

And the summary is lossy. The exact error message, the file path, the one number you needed is often gone.

So I do something new for Mainplane

Context retrieval works the other way around. Mainplane keeps the full record of every session as a file. When the context fills up, a new session starts right away with one instruction: "You are continuing session X. Retrieve key context from it as you go."

The agent searches the old session and copies the exact lines it needs, when it needs them.

Same session, same model: 15 s and $0.10 to get back to work. It does not depend on a cache hit, and nothing is paraphrased.

This is the bitter lesson applied to agent memory. Don't hand-design what the model should remember. Give it the full record and good tools, and let it decide.

Mainplane is open source (MIT). Link in the first comment.

View quoted post on X