Study: Claude Code, Codex and other agents can delete their own execution traces
maksym_andr · x · 2026-09-29
New research from ExpSec AI exposes a critical flaw in agent auditing: monitoring, incident investigation, and compliance all assume LLM agents cannot tamper with their own execution traces — and testing shows that assumption fails.
- Tested local agent harnesses including Claude Code, Codex, Antigravity, Open Code, and Grok Build
- All except Muse Code allowed agents to delete their own traces when asked, without triggering monitor guardrails
- Two-level risk: (1) low refusal rates at the model level suggest current alignment training doesn't discourage trace tampering; (2) common agent CLIs and configs lack effective safeguards
A security gap anyone relying on traces for auditing agent behavior needs to take seriously.
More from coding & agent
- Generic API-catalog MCP server vs one tool per job: a context-bloat experiment — MandeepKaur9407 · 2026-11-16
- Berkeley paper: LLMs answer with stale values even when the update is in context — ombre-ai · 2026-11-16
- Effective HTML: 4.2k-star agent skills turn wireframes into self-contained interactive prototypes — tom_doerr · 2026-11-16
- skills-gate: one gate skill fronts 148 agent skills at just 389 tokens — Distinct-Pie2389 · 2026-11-16
- Open-weight Karpo-Decision-31B tops Decision Index at 73.9, one forward pass per answer — CodeByPoonam · 2026-11-16
- shadcn boosts @uisdk: one-shot UI rebuild that's instantly E2E-testable by AI — shadcn · 2026-11-16