docs: add Latest News section to README

[vitali87]

Summary

• Add a "Latest News" section to the top of the README highlighting the three major recent features:
  1. Smart LOC Bounds (min/max across all backends)
  2. LLM Refinement Loop (iterative violation fixing)
  3. Auto-derived Minimum LOC (sensible defaults)

Test plan

• Verify rendering on GitHub

[greptile-apps]

Greptile Summary

This PR adds a "Latest News 🔥" section near the top of the README (between the badge block and the "Why pr-split?" section) to surface three recently shipped features to new visitors.

• Smart LOC Bounds: documents --min-loc / --max-loc flags enforced across all three partitioning backends (LLM, graph, CP-SAT)
• LLM Refinement Loop: documents --max-refinement-iterations for iterative violation fixing
• Auto-derived Minimum LOC: documents the automatic max_loc / 4 lower-bound fallback when --min-loc is not set

The change is purely additive documentation; no code is modified. The prose is clear, the placement is appropriate, and the formatting renders cleanly as standard GitHub Markdown.

Confidence Score: 5/5

Documentation-only change; safe to merge immediately.

The PR touches only README.md with a purely additive documentation section. No code, no configuration, no logic is changed. Prose is accurate, well-placed, and renders correctly in GitHub Markdown. No issues found that would warrant blocking merge.

No files require special attention.

Important Files Changed

Filename	Overview
README.md	Adds a "Latest News 🔥" section above "Why pr-split?" documenting three new features: Smart LOC Bounds, LLM Refinement Loop, and Auto-derived Minimum LOC — no issues found.

Flowchart

%%{init: {'theme': 'neutral'}}%%
flowchart TD
    A[pr-split: split PR] --> B{Partition backend}
    B --> C[LLM]
    B --> D[Graph]
    B --> E[CP-SAT]
    C & D & E --> F{LOC bounds check\n--min-loc / --max-loc}
    F -->|Undersized groups| G[Merge groups]
    F -->|Oversized groups| H[Penalise / split]
    F -->|Within bounds| I[Output split plan]
    G & H --> J{--max-refinement-iterations set?}
    J -->|Yes| K[Feed violations back to LLM\nRefinement loop]
    K --> F
    J -->|No| I
    I --> L[Create sub-PRs as DAG]

Loading

_{Reviews (2): Last reviewed commit: "docs: add fire emoji to Latest News head..." | Re-trigger Greptile}

[gemini-code-assist]

Code Review

This pull request updates the README.md file to include a 'Latest News' section detailing new features such as smart LOC bounds, an LLM refinement loop, and auto-derived minimum LOC settings. I have reviewed the changes and suggest updating the spelling of 'penalised' to 'penalized' to maintain consistency with the project's American English style.

[gemini-code-assist]

For consistency with American English spelling used elsewhere in the project (e.g., normalize), it's recommended to use 'penalized' instead of 'penalised'.

Suggested change

	**Smart LOC Bounds**: Configure `--min-loc` and `--max-loc` to control sub-PR size. All three partitioning backends (LLM, graph, CP-SAT) now enforce both lower and upper bounds — undersized groups get merged, oversized groups get penalised.
	**Smart LOC Bounds**: Configure `--min-loc` and `--max-loc` to control sub-PR size. All three partitioning backends (LLM, graph, CP-SAT) now enforce both lower and upper bounds — undersized groups get merged, oversized groups get penalized.