Anaconda on deslopping AI prose: the structural tells in PR descriptions
anacondainc · x · 2026-11-04
Anaconda's blog post by Andrew Huang (Oct 15, 2026) explains how to spot and fix "AI slop" in agent-written PR descriptions:
- Word-level tells (robust, leveraging, seamlessly, empowers) are only the surface — swapping them leaves prose that still reads as AI.
- Structural problems matter more: "It's not just X, it's Y" argues against a claim nobody made; triple lists exist because three sounds complete, not because three benefits were measured; the closing sentence restates the opener in bigger words.
- Fix: delete everything that isn't information — e.g. "Adds memoization to the reactive pipeline."
- Trust is the stake: once a reviewer spots slop, they doubt even the carefully hand-written parts.
- The accompanying "deslop loop" diagram: give a voice example you wrote → generate → delete first (openers, closers, hedges) → scan with deslopscan.py → reread the whole piece → review and ship. Each rewrite brings new tics, so one pass is never enough.
More from coding & agent
- ThunderAgent merged into LLM-D: 2x agent serving throughput, TTFT from ~50s to 2-4s — simran_s_arora · 2026-11-04
- Agents pass the test but fail the job: the requirement gap and the model gap — bigdata · 2026-11-04
- Does that new prompting trick actually work, or just look better? — jimmykoppel · 2026-11-04
- Vercel launches Security Review in public beta, free until Nov 10 — cramforce · 2026-11-04
- QuickProof: formal verification for C with specs written as plain C tests — thomasahle · 2026-11-04
- Project Moah: Open-Source Memory Linker That Gives AI Agents a Brain — starcount_games · 2026-11-04