pai-opencode/.coderabbit.yaml at main · Steffen025/pai-opencode · 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
# CodeRabbit Configuration — PAI-OpenCode
# Free Tier (OSS/personal project)
# Docs: https://docs.coderabbit.ai/getting-started/configure-coderabbit

language: "en"

# Review profile: assertive = flags everything, chill = only critical issues
# For active AI development, "chill" prevents review noise on AI-generated code
reviews:
  profile: "chill"
  request_changes_workflow: false  # CodeRabbit flags but humans decide merge

  # Auto-review all PRs targeting dev or main
  auto_review:
    enabled: true
    drafts: false  # don't review draft PRs
    base_branches:
      - "dev"
      - "main"

  # Path-specific instructions for CodeRabbit
  path_instructions:
    - path: ".opencode/skills/**"
      instructions: |
        PAI Skills — Domain-specific sub-algorithms.
        Verify:
        - SKILL.md follows PAI v3.0 format (USE WHEN triggers, MANDATORY/OPTIONAL sections)
        - Tools/ directory has CLI tools for automation
        - Workflows/ directory has operational procedures
        - No hardcoded API keys in tool implementations
    - path: ".opencode/agents/**"
      instructions: |
        PAI Agent Definitions — Persona, traits, voice configuration.
        Verify:
        - Frontmatter has required fields (name, description, model)
        - First-person voice convention followed ("I can help" not "Jeremy can help")
        - Color codes in hex format for UI consistency
    - path: ".opencode/plugins/**"
      instructions: |
        OpenCode Plugins — Lifecycle automation.
        Verify:
        - Hooks follow OpenCode plugin API
        - No blocking operations in init hooks
        - Proper error handling with fallbacks
    - path: "Tools/**"
      instructions: |
        CLI Tools — Pure execution (no LLM inside).
        Verify:
        - Uses Bun runtime (package.json has "type": "module")
        - No AI/LLM calls inside tools (Opus orchestrates, tools execute)
        - Proper CLI argument parsing with validation
        - Exit codes follow UNIX conventions (0=success, non-zero=error)
    - path: "skill-packs/**"
      instructions: |
        Modular Skill Packs — Optional PAI extensions.
        Verify:
        - Self-contained structure (SKILL.md + Agents/ + Workflows/)
        - Clear activation instructions
        - No dependencies on other skill-packs unless documented
    - path: "docs/**"
      instructions: |
        Documentation — Markdown with Obsidian conventions.
        Verify:
        - Frontmatter for Dataview queries where appropriate
        - ASCII diagrams for quick overview
        - Mermaid diagrams in collapsible sections for detail
        - Voice-safe text (no markdown in spoken summaries)
    - path: "**/*.md"
      instructions: |
        Markdown files — Obsidian-formatted documentation.
        Verify:
        - Callouts use `> [!type]` syntax
        - Frontmatter for important documents
        - German language: Use generic masculine (Teilnehmer, nicht Teilnehmer*innen)
        - Tables for structured comparisons

  # What to ignore in reviews
  path_filters:
    - "!node_modules/**"
    - "!bun.lock"
    - "!**/*.lock"
    - "!dist/**"
    - "!coverage/**"
    - "!.git/**"

# Chat configuration (when you @coderabbit in comments)
chat:
  auto_reply: true

# Code generation (for future use)
code_generation:
  docstrings:
    language: "en"