Source Intelligence
Follow Discord

DisclaimerUnofficial, and not affiliated with Anthropic. Nearly all of this is read straight out of what ships: npm bundles, captured prompts, published docs. Anthropic's own notes go in verbatim, marked as theirs. The rest is my reading, and every entry carries the strings behind it. If one looks wrong, vote it down and say why.

← All of v2.1.232 Home All releases ←olderv2.1.231 v2.1.233newer→
Claude Code v2.1.232

Saved prompt-cache diagnostics are bounds-checked when loaded

Under the hood
Useful1 Signal1
Elsewhere

Saved prompt-cache diagnostics are size-checked on load so a corrupt file is rejected.

What

The stored per-source prompt-cache diagnosis data is now validated on load with hard size limits, so a corrupted or oversized file is rejected instead of being loaded wholesale.

Details
  • Strings are capped at 200 characters, tool-name and per-tool hash maps at 2000 entries, block hash arrays at 5000, and message hashes at 20000.
  • The globalCacheStrategy field is checked against a fixed set of accepted values, one of which is tool_based.
  • Clearing is now selective: a subset of sources can be invalidated rather than wiping the entire map.
Evidence

tool_based

Strings lifted out of the shipped bundle, so the claim above can be checked against them.

See this entry in the whole of v2.1.232 →