ql-deslop

ql-deslop — post-review AI-slop cleanup

Purpose

Review approval does not mean the code is clean. LLM-authored code routinely introduces:

• Duplication of helpers / types / imports under slightly different names.
• Dead code — functions added but never called; exports that no consumer uses.
• Needless abstraction — factory for one caller; wrapper that only aliases; configurable field never varied.
• Boundary violations — private functions accessed via string names; cross-module reaches that skip the public API.
• Missing tests — paths and branches not exercised by any test.

Per Schreiber & Tippe 2025, 12.1% of AI-generated files contain ≥1 CWE, and per Zhong et al. 2026 (278k conversations) AI suggestions increase cyclomatic complexity 10-50× more than human edits. Without an explicit cleanup step, every autonomous run accumulates slop.

ql-deslop adds an explicit cleanup pass AFTER the story's review gate passes but BEFORE the merge to the feature branch. Borrowed directly from OMC Ralph Steps 7.5 (deslop) + 7.6 (regression re-verify).

When to use

• Automatically, as a mandatory step between ql-review PASS and the git merge in ql-execute.
• Manually, at any point after a story is implemented, to clean up before review.
• After absorbing a foreign branch (CPC promotion, cherry-pick) — the absorbed work may carry slop.

Opt-out

Single flag: --no-deslop. Used only when:

• The story is a pure test addition (nothing to clean).
• The deslop pass itself was run earlier in the same wave on the same files.
• The user explicitly demands skip (rare; recorded in commit trailer Deslop: skipped | <reason>).

Scope — strict file-list only

The deslop pass operates on git diff --name-only BASE_SHA..HEAD_SHA — the EXACT files the story touched. It never broadens scope silently. Reaching into unchanged files is an anti-pattern and is explicitly forbidden.

Smell taxonomy — 5 categories

1. Duplication

Detectors:

• Identifier similarity across changed files: extract top-level symbol names; flag near-duplicates (e.g., parseRequest / parseReq / handleParse) via Levenshtein or canonical-alpha-rename + exact match (per HyClone arXiv:2508.01357).
• Function-body AST hash: for each function in the diff, compute tree-sitter AST hash after alpha-renaming; flag duplicates across files.
• Import redundancy: same module imported as different names in sibling files.