Kimi K3.1 Pro stumbles on a real agent coding task: ordering errors sink the build
QuixiAI · x · 2026-10-24
QuixiAI shares a hands-on eval of Kimi K3.1 Pro on a large codebase agent task, calling the result "Mixed":
- What it did right: resumed from a handoff without restarting; paused and asked when files changed under it; strengthened migrator fixtures with hex-macro and typed-array assertions, refactored the seam into two modules, and correctly diagnosed the dev build failure.
- Where it went wrong: (1) ordering — it enabled the D5 strict pointer rule and started a runtime-wide byte-pointer conversion before fixing the three red corpora, the pcre2 seam, and wasm32 bundles, turning a latent disagreement into an unfixable red build; (2) introduced seed-blocked cchar spellings in compiler sources that can't compile under seed builds; (3) created a dependency cycle breaking the stage chain, with 27 modified uncommitted files.
- Suggested fixes: revert the strict pointer rule and cchar spellings, commit assertions/fixtures separately, follow handoff §3 in order, and no subagents or parallel batches on one worktree.
More from coding & agent
- Daniel Miessler: code is the lossy output — the primary artifact is the fully articulated ideal in language — JensHonack · 2026-10-24
- Builder pairs Grok Bot for firmware with locally-run Hermes Agent inside a Stack-chan robot — alexcovo_eth · 2026-10-24
- 30% of Git modelled in Lean 4 with coding agents, 8,000+ theorems proven, 100+ potential bugs found — kfountou · 2026-10-24
- REA-Prompts hits 1,700+ GitHub stars in days: prompts that make coding agents reverse-engineer any software — Promptmethus · 2026-10-24
- Veteran's advice: stop relying on specs alone, just talk to your LLM — MoonL88537 · 2026-10-24
- Agent orchestration tip: skip goal, use a manager agent with scheduled polling — pvncher · 2026-10-24