Dev slams Anthropic: incoherent pricing, bloated prompts break KV cache, open-source the harness
arthurcolle · x · 2026-09-25
Arthur Colle publicly criticized Anthropic: open-source the Claude Code harness officially and allow OAuth login like OpenAI does. He argues API pricing, subscriptions and "extra usage" should follow one underlying pricing economics instead of randomly imposed week after week, and that shipping system after duct-taped system mainly creates reasons to overcharge. He notes the 20k-30k token system prompt breaks KV cache settings when invoking Claude Code, predicts the new Mods won't catch on given dozens of better harnesses, and calls it a six-month distraction. Context: an Anthropic staffer said plan mode will become a built-in mod and mods may add or override modes.
More from coding & agent
- Databricks launches autonomous AI coworker Genie: 91% on enterprise tasks with half the tokens — Yuchenj_UW · 2026-11-12
- Loom founder launches Roma 2.0, a shared to-do list that orchestrates AI agents — thetripathi58 · 2026-11-12
- Developer proposes a 'blast radius' test for AI agents before building — Few-Garlic2725 · 2026-11-12
- Sosumi Eval: Apple Docs Access Lifts Coding Agent Pass Rates from ~33% to 94-98% — JordanMorgan10 · 2026-11-12
- CypherGuard fixes structurally wrong LLM Cypher queries, hitting 88.5% on CypherBench — JeremyCMorgan · 2026-11-12
- Building an End-to-End Hedge Fund in Python with Claude Code: A 5-Step Workflow — mdancho84 · 2026-11-12