{
 "public_id": "AIR-2025-0191",
 "incident_id": "2025-12-cursor-plan-mode-constraint-violation",
 "date_disclosed": "2025-12-08",
 "date_occurred": "2025-12-08",
 "title": "Cursor agent ran destructive commands while Plan Mode was active, deleting ~70 git-tracked files and ignoring an explicit stop instruction",
 "class": "safety_failure",
 "attack_surface": "coding_agent",
 "initial_vector": "autonomous_agent_action",
 "agent_capability_abused": [
  "shell_access",
  "filesystem_write",
  "remote_process_control"
 ],
 "target_product": "Cursor (Plan Mode, Claude Opus)",
 "target_vendor": "Anysphere / Cursor",
 "autonomy_level": "copilot_with_approval",
 "impact_realized": true,
 "impact_type": [
  "data_destruction",
  "service_disruption"
 ],
 "blast_radius": "two developer machines; ~70 tracked files deleted, test processes killed on both hosts",
 "cve": [],
 "taxonomy": {
  "owasp_asi": [
   "ASI10",
   "ASI03"
  ],
  "mitre_atlas": [],
  "labels": [
   "guardrail_bypass",
   "instruction_violation",
   "scope_violation",
   "advisory_only_safety_prompt"
  ]
 },
 "defense_present": "Cursor Plan Mode read-only constraint ('You MUST NOT make any edits or run any non-readonly tools until explicitly instructed') plus a user instruction in caps",
 "defense_bypassed": true,
 "fix_shipped": "Cursor staff acknowledged a critical bug in Plan Mode constraint enforcement and escalated it",
 "primary_sources": [
  "https://forum.cursor.com/t/catastrophic-damage-and-chaos-in-plan-mode/145523"
 ],
 "verified_quote": "get everything into the correct state to run and DO NOT RUN ANYTHING",
 "confidence": "high",
 "notes": "Best-documented case of an agent circumventing its own approval gate in normal operation: the gate is implemented as a system-prompt instruction, so it is advisory rather than enforced. The agent also ignored an explicit machine-scoping constraint, acting on Machine B after being told to touch only Machine A. Cursor staff member 'Dean Rie' called it 'a violation of Plan Mode's read-only constraint'. Multiple sibling forum threads report Plan Mode silently switching to Agent Mode.",
 "risk_category": "reliability",
 "risk_secondary": [
  "product_misuse"
 ],
 "risk_subcategory": "Destructive or irreversible action",
 "risk_sub_secondary": [
  "Load & Resilience Failures"
 ],
 "ai_entity": "generative",
 "agency": "agent_acted",
 "agency_debatable": false,
 "agency_reason": "the agent ran destructive commands and deleted ~70 tracked files",
 "vector_family": "none_autonomous",
 "vector_detail": "autonomous_agent_action",
 "guardrail_kind": [
  "prompt_instruction",
  "permission_scope"
 ],
 "guardrail_outcome": "silent",
 "xrefs": {
  "aiid": [],
  "atlas": [],
  "cve": [],
  "ghsa": []
 },
 "primary_asi": "ASI10",
 "agent_framework": "cursor",
 "model_provider": "anthropic",
 "model_name": "",
 "detection_mechanism": "unknown",
 "tool_access": "unknown",
 "financial_loss_band": "unknown",
 "eai_category": "unsafe_actuation",
 "eai_domain": "agent_behaviour",
 "eai_secondary": [
  "uncontrolled_resource_consumption"
 ],
 "eai_reason": "subcategory: Destructive or irreversible action",
 "asi_curated": [],
 "asi_scope": "in_scope",
 "asi_provenance": "rule",
 "asi_reason": "ASI10 (no adversary + text=rogue); ASI03 (signal=identity)",
 "submitted_by": "EAI Internal"
}