← ClaudeAtlas

resloplisted

Rewrite docstrings and comments by reading the actual implementation, call sites, and tests — produce grounded, factual prose instead of deleting slop. Sibling of /techne:deslop. Use when the user wants to replace overhyped or hallucinated documentation with accurate one- or two-line descriptions derived from what the code actually does.
ajbarea/techne · ★ 1 · AI & Automation · score 70
Install: claude install-skill ajbarea/techne
# Reslop Generative sibling of `/techne:deslop`. Where `/techne:deslop` deletes low-value prose, `/techne:reslop` *rewrites* it — grounded in the code, not in marketing instincts. On a path-scoped `/techne:deslop <path>` run, deslop auto-routes claim-bearing docstrings here; on a bare whole-repo sweep it lists them for you to run this directly. ## Repo context ```! cat .claude/skill-context.md 2>/dev/null || echo "(no .claude/skill-context.md — rewrites will use conservative defaults for grounding numeric claims)" ``` The injected `## slop_ground_truth` section (if present) names the files or docs that ground quantitative claims for this repo. Never introduce a numeric claim that can't trace to one of those sources. ## Hard rules 1. **Never invent behavior.** If the code doesn't show it, don't claim it. No "thread-safe" unless the code actually synchronizes. No "idempotent" unless it demonstrably is. No "future-proof" ever. 2. **Never invent performance claims.** Rewrites must not introduce numeric speedups, throughput figures, latency bounds, or scale claims unless they are cited in the repo's `slop_ground_truth` (per injected context). If the original prose said "10× faster" and you can't find a measurement that supports it, strike the claim — don't rephrase it. The "Unsupported quantitative / comparative claims" section of `${CLAUDE_PLUGIN_ROOT}/_shared/hate-words.md` is the pattern list. 3. **No hate-words.** See `${CLAUDE_PLUGIN_ROOT}/_shared/hate-words.md` — the r