seedgen 0.3.0

Zero-config database seed data generator with deterministic, constraint-safe fake data
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
84
85
86
87
88
89
90
91
92
93
94
95
96
97

# CodeRabbit Configuration for SeedGen
# Docs: https://docs.coderabbit.ai/getting-started/yaml-configuration
#
# INSTALL FIRST before this file does anything:
# 1. Open https://coderabbit.ai → Login with GitHub
# 2. Authorize → select the SeedGen repo
# 3. Done. This file is picked up automatically by CodeRabbit.

language: en

reviews:
  auto_review:
    enabled: true
    # Skip chore/release-only PRs
    ignore_title_keywords:
      - "chore"
      - "release"
    # Don't review non-code files
    path_filters:
      - "!**/*.md"
      - "!scenarios/*.yaml"
      - "!.github/**"
      - "!.gitignore"
      - "!LICENSE"

  # assertive = strict review, suitable for a project with strict invariants
  profile: assertive

  # Project-specific review rules per path
  path_instructions:
    - path: "src/**/*.rs"
      instructions: |
        This is a Rust project (SeedGen) with strict invariants. Review for:

        CRITICAL (must block PR if violated):
        1. DETERMINISM: Never use rand::thread_rng(), SystemTime::now() for randomness, or any non-deterministic source in library code. Only ChaCha8Rng seeded with u64 is allowed for all randomness.
        2. NO UNWRAP: Never use .unwrap() or .expect() in src/ library code. Always propagate errors with ? operator. Only src/cli/ and src/main.rs may use anyhow.
        3. SQL SAFETY: Never use format!() or string interpolation to build SQL values. Always use sqlx parameterized queries with bind().
        4. NO CREDENTIALS: Never log, print, or include database connection URLs in error messages, tracing output, or MCP responses. Always redact.

        IMPORTANT (should flag):
        5. FK INTEGRITY: Generated FK values must always reference existing parent rows.
        6. UNIQUE COMPLIANCE: UNIQUE columns must track seen values with HashSet + retry on collision.
        7. Use tracing macros (info!, debug!, warn!, error!) — never println!() or eprintln!() in library code.
        8. Imports ordered: std → external crates → internal modules, separated by blank lines.
        9. Error types use thiserror with descriptive messages.
        10. No Box<dyn Error> in library code — use typed errors.

    - path: "src/generators/**/*.rs"
      instructions: |
        Generator-specific rules:
        - All generators MUST accept &mut ChaCha8Rng (never thread_rng)
        - All generators MUST return the Value enum
        - Same seed MUST produce same output — determinism is non-negotiable
        - Each generator MUST have unit tests proving determinism (generate twice with same seed, assert equal)

    - path: "src/mcp/**/*.rs"
      instructions: |
        MCP server rules:
        - NO AI/LLM logic inside the MCP server — pure deterministic tool execution only
        - Production safety: REFUSE operations if database name contains "prod" or "production" (unless explicit override)
        - Connection URLs MUST be redacted from ALL responses and error messages
        - Timeout: 60 seconds per operation maximum
        - Transport: stdio (default) or HTTP+SSE

    - path: "src/resolver/**/*.rs"
      instructions: |
        Resolver rules:
        - Topological sort uses Kahn's algorithm
        - Self-referencing FKs must NOT create dependency edges
        - Cycle detection uses DFS with coloring (white/gray/black)
        - Cycles are resolved by breaking nullable or deferrable FK edges

    - path: "src/output/**/*.rs"
      instructions: |
        Output rules:
        - Batch size: 1000 rows per INSERT statement
        - Wrap in transaction (BEGIN/COMMIT)
        - SQL file output must include session_replication_role toggle
        - NEVER hold all generated rows in memory — stream in batches

    - path: "tests/**/*.rs"
      instructions: |
        Test rules:
        - Test naming: test_<module>_<behavior> (e.g., test_resolver_handles_diamond_dependency)
        - Clear arrange/act/assert structure
        - Integration tests gated with #[cfg(feature = "integration")]
        - Property-based tests gated with #[cfg(feature = "proptest")]

  tools:
    clippy:
      enabled: true
    shellcheck:
      enabled: true

chat:
  auto_reply: true