Month: January 2026
-
7 – LLM “Plan Fatigue”: Why Late Validation Steps Got Skipped (and a Fix That Worked for Me)
I’ve been building a rule-heavy review agent for public-API governance changes. Early phases (discovery, candidate generation) work reliably. Later phases—especially rigorous validation against the full rule set—frequently get ignored, skimmed, or half-done, even with explicit, repeated instructions to treat them as mandatory. This feels like a classic symptom: the model excels at creative/open-ended early work…
-
6 – The Pungent Stink of Gravity Repellents: Battling LLM Reversion in Code
You’ve fixed the same brittle code three times. The large language model (LLM) — the AI behind tools like ChatGPT, Claude, or Windsurf — accepts your robust change, such as using a regex to match any protocol in a URL: re.match(r’^([a-zA-Z][a-zA-Z0-9+.-]*):’, url). Then you ask for a small tweak to a nearby line. The model immediately…
-
5-Your Flailing AI Is Stuck on the “Novelty Cliff” and You’ll Never Fix It
I’ve been tuning a prompt for weeks. The rules are clearly stated. The AI model, or LLM, keeps misapplying them—cheerfully, confidently, never quite right. The tell: output is logically incoherent where it’s usually smoothly joined. More explanation doesn’t help. The LLM never says “this is new to me”—it just flails, always misunderstanding, always poorly executed…