rx4 0.7.0

The agent harness engine — loop, tools, providers, sessions, permissions, computer-use
Documentation 
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

{
  "dataset": "princeton-nlp/SWE-bench_Verified",
  "metric": "resolve_rate",
  "default_n": 10,
  "default_sample_seed": 0,
  "full_n": 500,
  "harnesses": [
    "codex",
    "pi",
    "tk",
    "omp",
    "fx",
    "opencode"
  ],
  "notes": {
    "source": "Detected ~/.codex/config.toml model=gpt-5.6-sol effort=low; second slot gpt-5.6-terra from ~/.codex/models_cache.json (priority 2). Same model+effort on every harness.",
    "tk_effort": "tk exec --effort is passed through when present; current telekinesis-rx4-consume tk exec only documents --model (no --effort yet).",
    "fx": "fx (vercel-labs/fx) non-interactive: fx ask --yolo --no-save; stdin /dev/null. Model via --model if advertised else FX_MODEL. Effort via --effort/--thinking if advertised. Unauthenticated (no fx login / setup / AI_GATEWAY_API_KEY) fail closed.",
    "omp": "omp v18 non-interactive: -p --print --cwd TASK --no-session --provider openai-codex --model openai-codex/<id> --thinking <effort>; stdin /dev/null. Bare gpt-5.6-sol fuzzy-matches cursor first (no CURSOR_ACCESS_TOKEN).",
    "opencode": "opencode non-interactive: opencode run --auto --dir TASK -m <id>; stdin /dev/null. Auth via ~/.local/share/opencode/auth.json (adapter never prints keys). deepseek-v4-flash uses cline-pass/cline-pass/deepseek-v4-flash."
  },
  "models": [
    {
      "id": "gpt-5.6-sol",
      "effort": "low",
      "codex": {
        "model": "gpt-5.6-sol",
        "config": "model_reasoning_effort=low"
      },
      "pi": {
        "provider": "openai-codex",
        "model": "gpt-5.6-sol",
        "thinking": "low"
      },
      "tk": {
        "model": "gpt-5.6-sol",
        "effort": "low"
      },
      "omp": {
        "model": "gpt-5.6-sol",
        "thinking": "low",
        "provider": "openai-codex"
      },
      "fx": {
        "model": "gpt-5.6-sol",
        "effort": "low"
      }
    },
    {
      "id": "gpt-5.6-terra",
      "effort": "low",
      "codex": {
        "model": "gpt-5.6-terra",
        "config": "model_reasoning_effort=low"
      },
      "pi": {
        "provider": "openai-codex",
        "model": "gpt-5.6-terra",
        "thinking": "low"
      },
      "tk": {
        "model": "gpt-5.6-terra",
        "effort": "low"
      },
      "omp": {
        "model": "gpt-5.6-terra",
        "thinking": "low",
        "provider": "openai-codex"
      },
      "fx": {
        "model": "gpt-5.6-terra",
        "effort": "low"
      }
    },
    {
      "id": "deepseek-v4-flash",
      "effort": "low",
      "opencode": {
        "model": "cline-pass/cline-pass/deepseek-v4-flash"
      }
    }
  ]
}