perf(kaos): stream readLines to avoid loading whole file into memory

[7Sageer]

Related Issue

No related issue. The problem is explained below.

Problem

Reading a large text file with the Read tool (including tail mode via a negative line_offset) loads the entire file into memory. LocalKaos.readLines used readFile + split('\n'), which holds the whole file as a Buffer, a decoded string, and a split array at the same time. On a 100 MB file this peaks at ~330 MB of resident memory even when only the last 100 lines are needed, and grows linearly with file size.

What changed

LocalKaos.readLines now reads the file in 64 KiB chunks and splits on the LF byte, decoding one line at a time. Peak memory is bounded by the chunk size plus the longest line rather than the whole file (~57 MB for the same 100 MB file, most of which is the Node.js baseline).

Behavior is unchanged: readTail / readForward keep their ring buffer, the Total lines in file report, and the MAX_LINES / MAX_BYTES caps. CRLF, lone CR, empty files, and the strict / replace / ignore decode modes are preserved. Added tests for line-ending fidelity, multi-chunk files, multibyte characters across the chunk boundary, and the three decode modes.

Checklist

• I have read the CONTRIBUTING document.
• I have linked a related issue, or explained the problem above.
• I have added tests that prove my feature works.
• Ran gen-changesets skill, or this PR needs no changeset.
• Ran gen-docs skill, or this PR needs no doc update.

[changeset-bot]

🦋 Changeset detected

Latest commit: f6d584e

The changes in this PR will be included in the next version bump.

This PR includes changesets to release 1 package

Name	Type
@moonshot-ai/kimi-code	Patch

Not sure what this means? Click here to learn what changesets are.

Click here if you're a maintainer who wants to add another changeset to this PR

[pkg-pr-new]

pnpm dlx https://pkg.pr.new/@moonshot-ai/kimi-code@f6d584e

npx https://pkg.pr.new/@moonshot-ai/kimi-code@f6d584e

commit: f6d584e

[itxaiohanglover]

Good performance fix — streaming file reads in 64KB chunks instead of loading the entire file into memory is the right approach for large files. The pending buffer handling for partial lines is correct. Changeset included.