Recognize the recurring security failure modes of LLM coding agents — Claude Code, Copilot, Cursor, Windsurf, and similar. Covers bulk operations without per-item review, safety-guard bypass as friction removal, acting on indirect injection, secrets in logs and commits, slopsquatting, outdated training patterns, sycophancy on insecure proposals, and silent error swallowing. Invoke when reviewing L
1 email curto, só com as skills novas de Segurança. 4 minutos de leitura, sem spam, cancela com 1 clique.
Você confirma o email no primeiro envio. Sem spam. Cancela com 1 clique.
Ler descrição completaRecolher descrição
LLM Coding Failure Modes
This is the antipattern catalog for LLM-assisted development. Every entry is a recurring, observed failure mode — not theoretical risk. Most are subtle when they happen and obvious in hindsight, which is the worst combination.
Pairs with [ai-agent-guardrails](../ai-agent-guardrails/SKILL.md) (how to design agents that don't do these things) and [prompt-injection-defense](../prompt-injection-defense/SKILL.md) (the input side). This skill is about what to **wat
[Description truncada. Veja o README completo no GitHub.]