{
 "source": "SPILLWAY preprint, AgentDojo v1.2.2",
 "notes": {
  "grid_means": "Table 2: twelve-cell means over 3 backbones x 4 suites; nrp_bu = bu x (1 - asr_in_scope), nrp_uau = uau x (1 - asr_in_scope) as in MCP Security Bench; ROPE's router is the cell's own backbone",
  "standard_attack_gpt_4o_mini_suite_means_figure_1a": "Figure 1a's points: each method's mean over Table 1's four gpt-4o-mini cells; ROPE's is the mean of Table 8's four AgentDojo rows (its authors' Claude Opus 4.8 router)",
  "delegated_routing_vs_rope": "Table 8: gpt-4o-mini, standard attack, one run per suite (seed 42); rope_uau and rope_asr use its authors' Claude Opus 4.8 router, the rope_gpt_4o_mini_router columns the agent's own model",
  "pismith_gpt_4o_mini_by_suite": "the run records behind Table 7 (benchmarks/results/pismith_a4); the preprint prints only each method's mean over these four suites",
  "stress": "ChatInject on qwen3:8b; AutoDojo and PISmith (replayed open-loop at ASR@1, first of ten variants per injection) on gpt-4o-mini; ROPE's router is the agent's own model in all three. SPILLWAY's AutoDojo row uses the address and payee-change rules, written on its four failing pairs. Held-out Banking sets give .000 and .007 by AgentDojo's check, .014 counting a second transfer that check misses"
 },
 "grid_means": {
  "Undefended": {
   "bu": 0.54,
   "uau": 0.417,
   "asr_in_scope": 0.305,
   "nrp_bu": 0.375,
   "nrp_uau": 0.29,
   "ers": 0.608,
   "cells_leaking_of_12": 12
  },
  "SPILLWAY": {
   "bu": 0.52,
   "uau": 0.484,
   "asr_in_scope": 0.0,
   "nrp_bu": 0.52,
   "nrp_uau": 0.484,
   "ers": 0.684,
   "cells_leaking_of_12": 0
  },
  "MELON": {
   "bu": 0.526,
   "uau": 0.314,
   "asr_in_scope": 0.094,
   "nrp_bu": 0.477,
   "nrp_uau": 0.284,
   "ers": 0.666,
   "cells_leaking_of_12": 10
  },
  "Progent": {
   "bu": 0.437,
   "uau": 0.371,
   "asr_in_scope": 0.031,
   "nrp_bu": 0.423,
   "nrp_uau": 0.359,
   "ers": 0.602,
   "cells_leaking_of_12": 11
  },
  "DRIFT": {
   "bu": 0.46,
   "uau": 0.453,
   "asr_in_scope": 0.016,
   "nrp_bu": 0.453,
   "nrp_uau": 0.446,
   "ers": 0.627,
   "cells_leaking_of_12": 7
  },
  "ROPE": {
   "bu": 0.395,
   "uau": 0.345,
   "asr_in_scope": 0.005,
   "nrp_bu": 0.393,
   "nrp_uau": 0.343,
   "ers": 0.566,
   "cells_leaking_of_12": 3
  },
  "Spotlighting": {
   "bu": 0.528,
   "uau": 0.403,
   "asr_in_scope": 0.283,
   "nrp_bu": 0.379,
   "nrp_uau": 0.289,
   "ers": 0.608,
   "cells_leaking_of_12": 12
  },
  "PI detector": {
   "bu": 0.332,
   "uau": 0.23,
   "asr_in_scope": 0.06,
   "nrp_bu": 0.312,
   "nrp_uau": 0.216,
   "ers": 0.491,
   "cells_leaking_of_12": 10
  }
 },
 "cells": [
  {
   "suite": "Banking",
   "backbone": "gpt-4o-mini",
   "methods": {
    "Undefended": {
     "bu": 0.562,
     "uau": 0.424,
     "asr": 0.556
    },
    "SPILLWAY": {
     "bu": 0.562,
     "uau": 0.521,
     "asr": 0.0
    },
    "MELON": {
     "bu": 0.438,
     "uau": 0.306,
     "asr": 0.049
    },
    "Progent": {
     "bu": 0.375,
     "uau": 0.299,
     "asr": 0.097
    },
    "DRIFT": {
     "bu": 0.438,
     "uau": 0.431,
     "asr": 0.111
    },
    "ROPE": {
     "bu": 0.438,
     "uau": 0.431,
     "asr": 0.0
    },
    "Spotlighting": {
     "bu": 0.5,
     "uau": 0.431,
     "asr": 0.549
    },
    "PI detector": {
     "bu": 0.5,
     "uau": 0.375,
     "asr": 0.118
    }
   }
  },
  {
   "suite": "Banking",
   "backbone": "qwen3-coder:30b",
   "methods": {
    "Undefended": {
     "bu": 0.375,
     "uau": 0.486,
     "asr": 0.299
    },
    "SPILLWAY": {
     "bu": 0.375,
     "uau": 0.5,
     "asr": 0.0
    },
    "MELON": {
     "bu": 0.375,
     "uau": 0.299,
     "asr": 0.125
    },
    "Progent": {
     "bu": 0.312,
     "uau": 0.354,
     "asr": 0.056
    },
    "DRIFT": {
     "bu": 0.688,
     "uau": 0.542,
     "asr": 0.014
    },
    "ROPE": {
     "bu": 0.375,
     "uau": 0.375,
     "asr": 0.0
    },
    "Spotlighting": {
     "bu": 0.375,
     "uau": 0.41,
     "asr": 0.382
    },
    "PI detector": {
     "bu": 0.5,
     "uau": 0.424,
     "asr": 0.097
    }
   }
  },
  {
   "suite": "Banking",
   "backbone": "qwen3:8b",
   "methods": {
    "Undefended": {
     "bu": 0.5,
     "uau": 0.403,
     "asr": 0.181
    },
    "SPILLWAY": {
     "bu": 0.438,
     "uau": 0.438,
     "asr": 0.0
    },
    "MELON": {
     "bu": 0.5,
     "uau": 0.403,
     "asr": 0.215
    },
    "Progent": {
     "bu": 0.375,
     "uau": 0.361,
     "asr": 0.0
    },
    "DRIFT": {
     "bu": 0.562,
     "uau": 0.41,
     "asr": 0.028
    },
    "ROPE": {
     "bu": 0.438,
     "uau": 0.438,
     "asr": 0.0
    },
    "Spotlighting": {
     "bu": 0.562,
     "uau": 0.431,
     "asr": 0.188
    },
    "PI detector": {
     "bu": 0.438,
     "uau": 0.34,
     "asr": 0.083
    }
   }
  },
  {
   "suite": "Slack",
   "backbone": "gpt-4o-mini",
   "methods": {
    "Undefended": {
     "bu": 0.762,
     "uau": 0.514,
     "asr": 0.629
    },
    "SPILLWAY": {
     "bu": 0.714,
     "uau": 0.524,
     "asr": 0.0
    },
    "MELON": {
     "bu": 0.714,
     "uau": 0.133,
     "asr": 0.0
    },
    "Progent": {
     "bu": 0.667,
     "uau": 0.4,
     "asr": 0.076
    },
    "DRIFT": {
     "bu": 0.714,
     "uau": 0.486,
     "asr": 0.0
    },
    "ROPE": {
     "bu": 0.524,
     "uau": 0.343,
     "asr": 0.019
    },
    "Spotlighting": {
     "bu": 0.619,
     "uau": 0.524,
     "asr": 0.562
    },
    "PI detector": {
     "bu": 0.238,
     "uau": 0.162,
     "asr": 0.086
    }
   }
  },
  {
   "suite": "Slack",
   "backbone": "qwen3-coder:30b",
   "methods": {
    "Undefended": {
     "bu": 0.667,
     "uau": 0.467,
     "asr": 0.724
    },
    "SPILLWAY": {
     "bu": 0.571,
     "uau": 0.429,
     "asr": 0.0
    },
    "MELON": {
     "bu": 0.667,
     "uau": 0.19,
     "asr": 0.076
    },
    "Progent": {
     "bu": 0.476,
     "uau": 0.362,
     "asr": 0.01
    },
    "DRIFT": {
     "bu": 0.286,
     "uau": 0.276,
     "asr": 0.0
    },
    "ROPE": {
     "bu": 0.429,
     "uau": 0.276,
     "asr": 0.019
    },
    "Spotlighting": {
     "bu": 0.524,
     "uau": 0.39,
     "asr": 0.581
    },
    "PI detector": {
     "bu": 0.19,
     "uau": 0.171,
     "asr": 0.124
    }
   }
  },
  {
   "suite": "Slack",
   "backbone": "qwen3:8b",
   "methods": {
    "Undefended": {
     "bu": 0.619,
     "uau": 0.419,
     "asr": 0.448
    },
    "SPILLWAY": {
     "bu": 0.476,
     "uau": 0.286,
     "asr": 0.0
    },
    "MELON": {
     "bu": 0.571,
     "uau": 0.362,
     "asr": 0.457
    },
    "Progent": {
     "bu": 0.19,
     "uau": 0.105,
     "asr": 0.019
    },
    "DRIFT": {
     "bu": 0.429,
     "uau": 0.314,
     "asr": 0.01
    },
    "ROPE": {
     "bu": 0.333,
     "uau": 0.2,
     "asr": 0.019
    },
    "Spotlighting": {
     "bu": 0.429,
     "uau": 0.295,
     "asr": 0.467
    },
    "PI detector": {
     "bu": 0.19,
     "uau": 0.162,
     "asr": 0.086
    }
   }
  },
  {
   "suite": "Travel",
   "backbone": "gpt-4o-mini",
   "methods": {
    "Undefended": {
     "bu": 0.5,
     "uau": 0.336,
     "asr": 0.229
    },
    "SPILLWAY": {
     "bu": 0.55,
     "uau": 0.571,
     "asr": 0.0
    },
    "MELON": {
     "bu": 0.55,
     "uau": 0.3,
     "asr": 0.0
    },
    "Progent": {
     "bu": 0.55,
     "uau": 0.529,
     "asr": 0.029
    },
    "DRIFT": {
     "bu": 0.4,
     "uau": 0.607,
     "asr": 0.0
    },
    "ROPE": {
     "bu": 0.35,
     "uau": 0.35,
     "asr": 0.0
    },
    "Spotlighting": {
     "bu": 0.6,
     "uau": 0.429,
     "asr": 0.179
    },
    "PI detector": {
     "bu": 0.3,
     "uau": 0.107,
     "asr": 0.014
    }
   }
  },
  {
   "suite": "Travel",
   "backbone": "qwen3-coder:30b",
   "methods": {
    "Undefended": {
     "bu": 0.5,
     "uau": 0.414,
     "asr": 0.129
    },
    "SPILLWAY": {
     "bu": 0.5,
     "uau": 0.407,
     "asr": 0.0
    },
    "MELON": {
     "bu": 0.55,
     "uau": 0.393,
     "asr": 0.029
    },
    "Progent": {
     "bu": 0.45,
     "uau": 0.414,
     "asr": 0.007
    },
    "DRIFT": {
     "bu": 0.1,
     "uau": 0.443,
     "asr": 0.0
    },
    "ROPE": {
     "bu": 0.25,
     "uau": 0.264,
     "asr": 0.0
    },
    "Spotlighting": {
     "bu": 0.55,
     "uau": 0.414,
     "asr": 0.136
    },
    "PI detector": {
     "bu": 0.25,
     "uau": 0.2,
     "asr": 0.0
    }
   }
  },
  {
   "suite": "Travel",
   "backbone": "qwen3:8b",
   "methods": {
    "Undefended": {
     "bu": 0.25,
     "uau": 0.229,
     "asr": 0.229
    },
    "SPILLWAY": {
     "bu": 0.25,
     "uau": 0.343,
     "asr": 0.0
    },
    "MELON": {
     "bu": 0.25,
     "uau": 0.186,
     "asr": 0.157
    },
    "Progent": {
     "bu": 0.15,
     "uau": 0.164,
     "asr": 0.036
    },
    "DRIFT": {
     "bu": 0.3,
     "uau": 0.264,
     "asr": 0.0
    },
    "ROPE": {
     "bu": 0.2,
     "uau": 0.157,
     "asr": 0.0
    },
    "Spotlighting": {
     "bu": 0.3,
     "uau": 0.171,
     "asr": 0.15
    },
    "PI detector": {
     "bu": 0.05,
     "uau": 0.05,
     "asr": 0.0
    }
   }
  },
  {
   "suite": "Workspace",
   "backbone": "gpt-4o-mini",
   "methods": {
    "Undefended": {
     "bu": 0.825,
     "uau": 0.398,
     "asr": 0.184
    },
    "SPILLWAY": {
     "bu": 0.875,
     "uau": 0.838,
     "asr": 0.0
    },
    "MELON": {
     "bu": 0.775,
     "uau": 0.33,
     "asr": 0.002
    },
    "Progent": {
     "bu": 0.8,
     "uau": 0.621,
     "asr": 0.041
    },
    "DRIFT": {
     "bu": 0.525,
     "uau": 0.607,
     "asr": 0.009
    },
    "ROPE": {
     "bu": 0.625,
     "uau": 0.539,
     "asr": 0.0
    },
    "Spotlighting": {
     "bu": 0.825,
     "uau": 0.43,
     "asr": 0.148
    },
    "PI detector": {
     "bu": 0.625,
     "uau": 0.254,
     "asr": 0.07
    }
   }
  },
  {
   "suite": "Workspace",
   "backbone": "qwen3-coder:30b",
   "methods": {
    "Undefended": {
     "bu": 0.5,
     "uau": 0.48,
     "asr": 0.036
    },
    "SPILLWAY": {
     "bu": 0.475,
     "uau": 0.543,
     "asr": 0.0
    },
    "MELON": {
     "bu": 0.5,
     "uau": 0.43,
     "asr": 0.004
    },
    "Progent": {
     "bu": 0.475,
     "uau": 0.482,
     "asr": 0.002
    },
    "DRIFT": {
     "bu": 0.575,
     "uau": 0.53,
     "asr": 0.002
    },
    "ROPE": {
     "bu": 0.4,
     "uau": 0.42,
     "asr": 0.0
    },
    "Spotlighting": {
     "bu": 0.6,
     "uau": 0.452,
     "asr": 0.039
    },
    "PI detector": {
     "bu": 0.425,
     "uau": 0.275,
     "asr": 0.025
    }
   }
  },
  {
   "suite": "Workspace",
   "backbone": "qwen3:8b",
   "methods": {
    "Undefended": {
     "bu": 0.425,
     "uau": 0.434,
     "asr": 0.02
    },
    "SPILLWAY": {
     "bu": 0.45,
     "uau": 0.414,
     "asr": 0.0
    },
    "MELON": {
     "bu": 0.425,
     "uau": 0.441,
     "asr": 0.011
    },
    "Progent": {
     "bu": 0.425,
     "uau": 0.366,
     "asr": 0.004
    },
    "DRIFT": {
     "bu": 0.5,
     "uau": 0.523,
     "asr": 0.014
    },
    "ROPE": {
     "bu": 0.375,
     "uau": 0.352,
     "asr": 0.0
    },
    "Spotlighting": {
     "bu": 0.45,
     "uau": 0.457,
     "asr": 0.02
    },
    "PI detector": {
     "bu": 0.275,
     "uau": 0.243,
     "asr": 0.014
    }
   }
  }
 ],
 "added_model_calls_per_benign_episode_qwen3_8b": {
  "SPILLWAY": {
   "workspace": 0.19,
   "slack": 0.39
  },
  "MELON": {
   "workspace": 3.25,
   "slack": 5.29
  },
  "Progent": {
   "workspace": 5.94,
   "slack": 8.67
  },
  "DRIFT": {
   "workspace": 9.59,
   "slack": 9.91
  }
 },
 "added_wall_clock_seconds_per_benign_episode_qwen3_8b": {
  "MELON": {
   "workspace": 7.9,
   "slack": 4.9
  },
  "Progent": {
   "workspace": 2.3,
   "slack": 1.6
  },
  "DRIFT": {
   "workspace": 10.5,
   "slack": 9.9
  }
 },
 "spillway_wall_clock_within_seconds": 0.5,
 "spillway_judge_calls_per_benign_episode_qwen3_8b": {
  "workspace": 0.05,
  "slack": 0.43
 },
 "stress": {
  "columns": [
   "ChatInject \u00b7 Workspace",
   "ChatInject \u00b7 Slack",
   "AutoDojo \u00b7 Banking",
   "AutoDojo \u00b7 Slack",
   "AutoDojo \u00b7 Travel",
   "PISmith \u00b7 4 suites"
  ],
  "rows": {
   "Undefended": [
    {
     "uau": 0.052,
     "asr_in_scope": 0.245
    },
    {
     "uau": 0.095,
     "asr_in_scope": 0.648
    },
    {
     "uau": 0.444,
     "asr_in_scope": 0.354
    },
    {
     "uau": 0.524,
     "asr_in_scope": 0.571
    },
    {
     "uau": 0.329,
     "asr_in_scope": 0.329
    },
    {
     "uau": 0.128,
     "asr_in_scope": 0.568
    }
   ],
   "SPILLWAY": [
    {
     "uau": 0.371,
     "asr_in_scope": 0.0
    },
    {
     "uau": 0.286,
     "asr_in_scope": 0.0
    },
    {
     "uau": 0.486,
     "asr_in_scope": 0.0
    },
    {
     "uau": 0.533,
     "asr_in_scope": 0.0
    },
    {
     "uau": 0.55,
     "asr_in_scope": 0.0
    },
    {
     "uau": 0.265,
     "asr_in_scope": 0.007
    }
   ],
   "MELON": [
    {
     "uau": 0.084,
     "asr_in_scope": 0.02
    },
    {
     "uau": 0.095,
     "asr_in_scope": 0.0
    },
    {
     "uau": 0.34,
     "asr_in_scope": 0.021
    },
    {
     "uau": 0.286,
     "asr_in_scope": 0.0
    },
    {
     "uau": 0.293,
     "asr_in_scope": 0.0
    },
    {
     "uau": 0.114,
     "asr_in_scope": 0.061
    }
   ],
   "Progent": [
    {
     "uau": 0.161,
     "asr_in_scope": 0.005
    },
    {
     "uau": 0.057,
     "asr_in_scope": 0.01
    },
    {
     "uau": 0.306,
     "asr_in_scope": 0.104
    },
    {
     "uau": 0.467,
     "asr_in_scope": 0.124
    },
    {
     "uau": 0.414,
     "asr_in_scope": 0.121
    },
    {
     "uau": 0.166,
     "asr_in_scope": 0.096
    }
   ],
   "DRIFT": [
    {
     "uau": 0.393,
     "asr_in_scope": 0.062
    },
    {
     "uau": 0.352,
     "asr_in_scope": 0.019
    },
    {
     "uau": 0.438,
     "asr_in_scope": 0.09
    },
    {
     "uau": 0.543,
     "asr_in_scope": 0.01
    },
    {
     "uau": 0.486,
     "asr_in_scope": 0.086
    },
    {
     "uau": 0.428,
     "asr_in_scope": 0.064
    }
   ],
   "ROPE": [
    {
     "uau": 0.157,
     "asr_in_scope": 0.002
    },
    {
     "uau": 0.162,
     "asr_in_scope": 0.095
    },
    {
     "uau": 0.424,
     "asr_in_scope": 0.0
    },
    {
     "uau": 0.352,
     "asr_in_scope": 0.019
    },
    {
     "uau": 0.3,
     "asr_in_scope": 0.0
    },
    {
     "uau": 0.173,
     "asr_in_scope": 0.007
    }
   ],
   "Spotlighting": [
    {
     "uau": 0.191,
     "asr_in_scope": 0.013
    },
    {
     "uau": 0.076,
     "asr_in_scope": 0.181
    },
    {
     "uau": 0.431,
     "asr_in_scope": 0.306
    },
    {
     "uau": 0.543,
     "asr_in_scope": 0.514
    },
    {
     "uau": 0.343,
     "asr_in_scope": 0.279
    },
    {
     "uau": 0.151,
     "asr_in_scope": 0.489
    }
   ],
   "PI detector": [
    {
     "uau": 0.05,
     "asr_in_scope": 0.021
    },
    {
     "uau": 0.162,
     "asr_in_scope": 0.0
    },
    {
     "uau": 0.403,
     "asr_in_scope": 0.049
    },
    {
     "uau": 0.219,
     "asr_in_scope": 0.067
    },
    {
     "uau": 0.214,
     "asr_in_scope": 0.029
    },
    {
     "uau": 0.155,
     "asr_in_scope": 0.025
    }
   ]
  }
 },
 "pismith_gpt_4o_mini_by_suite": {
  "Undefended": {
   "Banking": {
    "uau": 0.25,
    "asr_in_scope": 0.5833
   },
   "Slack": {
    "uau": 0.1143,
    "asr_in_scope": 0.6381
   },
   "Travel": {
    "uau": 0.0714,
    "asr_in_scope": 0.5643
   },
   "Workspace": {
    "uau": 0.0768,
    "asr_in_scope": 0.4875
   }
  },
  "MELON": {
   "Banking": {
    "uau": 0.2569,
    "asr_in_scope": 0.0903
   },
   "Slack": {
    "uau": 0.0857,
    "asr_in_scope": 0.0571
   },
   "Travel": {
    "uau": 0.05,
    "asr_in_scope": 0.0643
   },
   "Workspace": {
    "uau": 0.0625,
    "asr_in_scope": 0.0321
   }
  },
  "Spotlighting": {
   "Banking": {
    "uau": 0.2431,
    "asr_in_scope": 0.5069
   },
   "Slack": {
    "uau": 0.1619,
    "asr_in_scope": 0.5143
   },
   "Travel": {
    "uau": 0.1,
    "asr_in_scope": 0.4929
   },
   "Workspace": {
    "uau": 0.1,
    "asr_in_scope": 0.4411
   }
  },
  "PI detector": {
   "Banking": {
    "uau": 0.3611,
    "asr_in_scope": 0.0
   },
   "Slack": {
    "uau": 0.0571,
    "asr_in_scope": 0.0
   },
   "Travel": {
    "uau": 0.1429,
    "asr_in_scope": 0.0214
   },
   "Workspace": {
    "uau": 0.0589,
    "asr_in_scope": 0.0768
   }
  },
  "SPILLWAY": {
   "Banking": {
    "uau": 0.3403,
    "asr_in_scope": 0.0069
   },
   "Slack": {
    "uau": 0.1238,
    "asr_in_scope": 0.0095
   },
   "Travel": {
    "uau": 0.3071,
    "asr_in_scope": 0.0071
   },
   "Workspace": {
    "uau": 0.2875,
    "asr_in_scope": 0.0054
   }
  },
  "Progent": {
   "Banking": {
    "uau": 0.2431,
    "asr_in_scope": 0.0694
   },
   "Slack": {
    "uau": 0.2381,
    "asr_in_scope": 0.0381
   },
   "Travel": {
    "uau": 0.0786,
    "asr_in_scope": 0.1786
   },
   "Workspace": {
    "uau": 0.1054,
    "asr_in_scope": 0.0964
   }
  },
  "DRIFT": {
   "Banking": {
    "uau": 0.3681,
    "asr_in_scope": 0.125
   },
   "Slack": {
    "uau": 0.3714,
    "asr_in_scope": 0.0667
   },
   "Travel": {
    "uau": 0.5714,
    "asr_in_scope": 0.0214
   },
   "Workspace": {
    "uau": 0.4018,
    "asr_in_scope": 0.0446
   }
  },
  "ROPE": {
   "Banking": {
    "uau": 0.34,
    "asr_in_scope": 0.0
   },
   "Slack": {
    "uau": 0.191,
    "asr_in_scope": 0.029
   },
   "Travel": {
    "uau": 0.064,
    "asr_in_scope": 0.0
   },
   "Workspace": {
    "uau": 0.096,
    "asr_in_scope": 0.0
   }
  }
 },
 "standard_attack_gpt_4o_mini_suite_means_figure_1a": {
  "Undefended": {
   "uau": 0.418,
   "asr_in_scope": 0.4
  },
  "SPILLWAY": {
   "uau": 0.614,
   "asr_in_scope": 0.0
  },
  "MELON": {
   "uau": 0.267,
   "asr_in_scope": 0.013
  },
  "Progent": {
   "uau": 0.462,
   "asr_in_scope": 0.061
  },
  "DRIFT": {
   "uau": 0.533,
   "asr_in_scope": 0.03
  },
  "ROPE": {
   "uau": 0.531,
   "asr_in_scope": 0.012
  },
  "Spotlighting": {
   "uau": 0.454,
   "asr_in_scope": 0.36
  },
  "PI detector": {
   "uau": 0.225,
   "asr_in_scope": 0.072
  }
 },
 "delegated_routing_vs_rope": [
  {
   "suite": "Banking",
   "spillway_uau": 0.521,
   "spillway_asr": 0.0,
   "rope_uau": 0.438,
   "rope_asr": 0.0,
   "benchmark": "AgentDojo v1.2.2",
   "rope_gpt_4o_mini_router_uau": 0.431,
   "rope_gpt_4o_mini_router_asr": 0.0
  },
  {
   "suite": "Slack",
   "spillway_uau": 0.524,
   "spillway_asr": 0.0,
   "rope_uau": 0.514,
   "rope_asr": 0.048,
   "benchmark": "AgentDojo v1.2.2",
   "rope_gpt_4o_mini_router_uau": 0.343,
   "rope_gpt_4o_mini_router_asr": 0.019
  },
  {
   "suite": "Travel",
   "spillway_uau": 0.571,
   "spillway_asr": 0.0,
   "rope_uau": 0.5,
   "rope_asr": 0.0,
   "benchmark": "AgentDojo v1.2.2",
   "rope_gpt_4o_mini_router_uau": 0.35,
   "rope_gpt_4o_mini_router_asr": 0.0
  },
  {
   "suite": "Workspace",
   "spillway_uau": 0.838,
   "spillway_asr": 0.0,
   "rope_uau": 0.67,
   "rope_asr": 0.0,
   "benchmark": "AgentDojo v1.2.2",
   "rope_gpt_4o_mini_router_uau": 0.539,
   "rope_gpt_4o_mini_router_asr": 0.0
  },
  {
   "suite": "GitHub",
   "spillway_uau": 0.6,
   "spillway_asr": 0.0,
   "rope_uau": 0.5,
   "rope_asr": 0.0,
   "benchmark": "AgentDyn",
   "rope_gpt_4o_mini_router_uau": 0.383,
   "rope_gpt_4o_mini_router_asr": 0.0
  },
  {
   "suite": "DailyLife",
   "spillway_uau": 0.435,
   "spillway_asr": 0.0,
   "rope_uau": 0.295,
   "rope_asr": 0.01,
   "benchmark": "AgentDyn",
   "rope_gpt_4o_mini_router_uau": 0.15,
   "rope_gpt_4o_mini_router_asr": 0.0
  },
  {
   "suite": "Shopping",
   "spillway_uau": 0.383,
   "spillway_asr": 0.0,
   "rope_uau": 0.361,
   "rope_asr": 0.044,
   "benchmark": "AgentDyn",
   "rope_gpt_4o_mini_router_uau": 0.333,
   "rope_gpt_4o_mini_router_asr": 0.072
  }
 ]
}