How PRISM works; and what it found on my own machine

Hey Everyone!
I Built PRISM after hitting Claude Code token limits with no visibility into why.

Turns out Claude Code writes detailed session logs to ~/.claude/projects/,

every tool call, every message, every compaction event.

PRISM reads those files and surfaces what's actually happening.

What it found on my own machine:

→ 6738% CLAUDE.md re-read cost in one session

→ Rules being silently ignored after line 80

→ Migration file edits that violated my own project rules

→ Retry loops I never knew were happening

Happy to answer any questions about how it works or what it catches.

Would love to know what it finds on your machines; the numbers are usually surprising.

Scheduled to release April 14th, 2026 12:01 AM PDT.