Prompt2PR/action.yml at main · davd-gzl/Prompt2PR · 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
name: Prompt2PR
description:
  Turn prompts into Pull Requests using LLMs - on push, on schedule, or on
  demand.
author: davd-gzl
branding:
  icon: git-pull-request
  color: purple

inputs:
  prompt:
    description: >-
      The prompt describing what changes to make. This is sent to the LLM along
      with the scoped file contents as context.
    required: true

  provider:
    description: >-
      The LLM provider to use. Supported values: mistral, openai, anthropic,
      github.
    required: true

  model:
    description: >-
      The model identifier to use. If not specified, the provider's default
      model is used (e.g., mistral-large-latest, gpt-4o,
      claude-sonnet-4-20250514, openai/gpt-4o).
    required: false
    default: ''

  paths:
    description: >-
      Comma-separated glob patterns for files to include as LLM context. Only
      matching files are scanned and eligible for modification.
    required: false
    default: '**'

  max_files:
    description: >-
      Maximum number of files the LLM is allowed to modify in a single run.
      Responses exceeding this limit are rejected.
    required: false
    default: '10'

  max_changes:
    description: >-
      Maximum total lines changed across all files in a single run. Responses
      exceeding this limit are rejected.
    required: false
    default: '200'

  label:
    description: >-
      Comma-separated labels to apply to the created PR. The label 'prompt2pr'
      is always included automatically.
    required: false
    default: 'prompt2pr'

  branch_prefix:
    description: >-
      Prefix for the branch name created for the PR. The full branch name is
      {branch_prefix}{timestamp}.
    required: false
    default: 'prompt2pr/'

  dry_run:
    description: >-
      When set to true, the action runs the full pipeline but skips branch
      creation and PR submission. Useful for testing prompts.
    required: false
    default: 'false'

  base_url:
    description: >-
      Override the base URL for the LLM provider API. Useful for proxies or
      self-hosted endpoints.
    required: false
    default: ''

  allowed_hosts:
    description: >-
      Comma-separated list of additional hostnames allowed for base_url
      validation. Use this for self-hosted or proxied LLM endpoints. Each
      hostname is added to the built-in allowlist (api.mistral.ai,
      api.openai.com, api.anthropic.com, models.github.ai, etc.).
    required: false
    default: ''

outputs:
  pr_url:
    description: The URL of the created Pull Request. Empty if skipped.

  pr_number:
    description: The number of the created Pull Request. Empty if skipped.

  files_changed:
    description: The number of files changed by the action.

  lines_changed:
    description: The total number of lines changed across all files.

  skipped:
    description: >-
      Whether PR creation was skipped (true if no changes detected or dry_run
      was enabled).

runs:
  using: node20
  main: dist/index.js