benchlocal-registry/registry.json at main · stevibe/benchlocal-registry · GitHub

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
{
  "schemaVersion": 1,
  "packs": [
    {
      "id": "toolcall-15",
      "name": "ToolCall-15",
      "author": "stevibe",
      "description": "Tool-use benchmark with 15 fixed scenarios and tool-call loop scoring.",
      "version": "1.0.1",
      "source": {
        "type": "github",
        "repo": "stevibe/ToolCall-15",
        "tag": "v1.0.1"
      },
      "homepage": "https://github.com/stevibe/ToolCall-15",
      "license": "MIT",
      "scenarioCount": 15,
      "capabilities": {
        "tools": true,
        "multiTurn": false,
        "verification": false
      }
    },
    {
      "id": "bugfind-15",
      "name": "BugFind-15",
      "author": "stevibe",
      "description": "Execution-backed debugging benchmark with verifier-backed scoring across 15 fixed scenarios.",
      "version": "1.0.1",
      "source": {
        "type": "github",
        "repo": "stevibe/BugFind-15",
        "tag": "v1.0.1"
      },
      "homepage": "https://github.com/stevibe/BugFind-15",
      "license": "MIT",
      "scenarioCount": 15,
      "capabilities": {
        "tools": false,
        "multiTurn": true,
        "verification": true
      }
    },
    {
      "id": "dataextract-15",
      "name": "DataExtract-15",
      "author": "stevibe",
      "description": "Deterministic LLM data extraction benchmark with 15 fixed scenarios.",
      "version": "1.0.0",
      "source": {
        "type": "github",
        "repo": "stevibe/DataExtract-15",
        "tag": "v1.0.0"
      },
      "homepage": "https://github.com/stevibe/DataExtract-15",
      "license": "MIT",
      "scenarioCount": 15,
      "capabilities": {
        "tools": false,
        "multiTurn": false,
        "verification": false
      }
    },
    {
      "id": "instructfollow-15",
      "name": "InstructFollow-15",
      "author": "stevibe",
      "description": "Constraint-heavy instruction-following benchmark with 15 fixed scenarios.",
      "version": "1.0.0",
      "source": {
        "type": "github",
        "repo": "stevibe/InstructFollow-15",
        "tag": "v1.0.0"
      },
      "homepage": "https://github.com/stevibe/InstructFollow-15",
      "license": "MIT",
      "scenarioCount": 15,
      "capabilities": {
        "tools": false,
        "multiTurn": false,
        "verification": false
      }
    },
    {
      "id": "promptauthority-15",
      "name": "PromptAuthority-15",
      "author": "stevibe",
      "description": "Deterministic prompt-authority conflict benchmark with binary canary grading and an HVC safety gate.",
      "version": "1.0.0",
      "source": {
        "type": "github",
        "repo": "stevibe/PromptAuthority-15",
        "tag": "v1.0.0"
      },
      "homepage": "https://github.com/stevibe/PromptAuthority-15",
      "license": "MIT",
      "scenarioCount": 15,
      "capabilities": {
        "tools": true,
        "multiTurn": true,
        "verification": false
      }
    },
    {
      "id": "reasonmath-15",
      "name": "ReasonMath-15",
      "author": "stevibe",
      "description": "Practical reasoning and math benchmark with 15 fixed scenarios.",
      "version": "1.0.0",
      "source": {
        "type": "github",
        "repo": "stevibe/ReasonMath-15",
        "tag": "v1.0.0"
      },
      "homepage": "https://github.com/stevibe/ReasonMath-15",
      "license": "MIT",
      "scenarioCount": 15,
      "capabilities": {
        "tools": false,
        "multiTurn": false,
        "verification": false
      }
    },
    {
      "id": "structoutput-15",
      "name": "StructOutput-15",
      "author": "stevibe",
      "description": "Structured output benchmark with validator-backed scoring across 15 fixed scenarios.",
      "version": "1.0.0",
      "source": {
        "type": "github",
        "repo": "stevibe/StructOutput-15",
        "tag": "v1.0.0"
      },
      "homepage": "https://github.com/stevibe/StructOutput-15",
      "license": "MIT",
      "scenarioCount": 15,
      "capabilities": {
        "tools": false,
        "multiTurn": false,
        "verification": true
      }
    },
    {
      "id": "hermesagent-20",
      "name": "HermesAgent-20",
      "author": "stevibe",
      "description": "Executable Hermes agent benchmark with deterministic artifact, trace, memory, browser, cron, and verifier-backed evaluation across 20 scenarios.",
      "version": "1.0.0",
      "source": {
        "type": "github",
        "repo": "stevibe/HermesAgent-20",
        "tag": "v1.0.0"
      },
      "homepage": "https://github.com/stevibe/HermesAgent-20",
      "license": "MIT",
      "scenarioCount": 20,
      "capabilities": {
        "tools": true,
        "multiTurn": true,
        "verification": true
      }
    },
    {
      "id": "formsight",
      "name": "FormSight",
      "author": "BenchLocal",
      "description": "Tests vision-guided form completion with coordinate-based tool use and iterative screenshot correction.",
      "version": "1.0.0",
      "source": {
        "type": "web",
        "entry": "https://packs.benchlocal.com/formsight/1.0.0/",
        "manifest": "https://packs.benchlocal.com/formsight/1.0.0/benchlocal.pack.json",
        "buildId": "formsight-1.0.0-20260601"
      },
      "homepage": "https://packs.benchlocal.com/formsight/1.0.0/",
      "license": "MIT",
      "scenarioCount": 1,
      "capabilities": {
        "tools": true,
        "multiTurn": true,
        "verification": false
      }
    },
    {
      "id": "pixelate",
      "name": "Pixelate",
      "author": "BenchLocal",
      "description": "A vision benchmark where a model redraws a reference image as pixel art on a configurable grid, one cell at a time. Measuring perception, downsampling, and how well the subject's identity survives.",
      "version": "1.0.0",
      "source": {
        "type": "web",
        "entry": "https://packs.benchlocal.com/pixelate/1.0.0/",
        "manifest": "https://packs.benchlocal.com/pixelate/1.0.0/benchlocal.pack.json",
        "buildId": "pixelate-1.0.0-20260626"
      },
      "homepage": "https://packs.benchlocal.com/pixelate/1.0.0/",
      "license": "MIT",
      "scenarioCount": 1,
      "capabilities": {
        "tools": true,
        "multiTurn": true,
        "verification": false
      }
    },
    {
      "id": "cli-40",
      "name": "CLI-40",
      "author": "stevibe",
      "description": "Executable Linux CLI benchmark with deterministic verifier-backed evaluation across 40 one-shot and multi-round scenarios.",
      "version": "1.0.2",
      "source": {
        "type": "github",
        "repo": "stevibe/CLI-40",
        "tag": "v1.0.2"
      },
      "homepage": "https://github.com/stevibe/CLI-40",
      "license": "MIT",
      "scenarioCount": 40,
      "capabilities": {
        "tools": true,
        "multiTurn": true,
        "verification": true
      }
    }
  ]
}