.coderabbit.yml

# yaml-language-server: $schema=https://coderabbit.ai/integrations/schema.v2.json
knowledge_base:
  # Link the MCP server repo so reviews can catch cross-repo breaking changes.
  # Schema allows maxItems: 1.
  linked_repositories:
    - repository: elastic/semantic-code-search-mcp-server
      instructions: |
        This indexer and the MCP server are coupled.
        When PRs change index mappings, document shape, APIs, or auth/config semantics, cross-check the MCP server expectations and CALL OUT any potential incompatibilities or breaking changes in the review/summary.
        Do not make changes in the MCP server repo as part of this PR; treat it as a reference only.

reviews:
  inheritance: true
  profile: assertive
  request_changes_workflow: true

  high_level_summary: false
  high_level_summary_in_walkthrough: false

  review_status: false
  review_details: false
  commit_status: true
  fail_commit_status: false

  collapse_walkthrough: true
  changed_files_summary: false
  sequence_diagrams: true
  estimate_code_review_effort: false
  assess_linked_issues: true
  related_issues: true
  related_prs: true
  suggested_labels: false
  auto_apply_labels: false
  suggested_reviewers: false
  auto_assign_reviewers: false

  in_progress_fortune: false
  poem: false

  enable_prompt_for_ai_agents: false

  # Keep reviews focused on authored source, not generated/large vendor dirs.
  path_filters:
    - '!dist/**'
    - '!node_modules/**'
    - '!.repos/**'
    - '!.queues/**'
    - '!coverage/**'
    - '!**/*.min.*'
    - '!**/*.map'

  path_instructions:
    - path: 'src/**/*.ts'
      instructions: |
        Focus on correctness, determinism, and performance under high-throughput indexing.
        Verify:
        - Robust error handling and actionable errors (include enough context, avoid noisy logs)
        - Concurrency safety (queues, worker pools, retries, cancellation)
        - Elasticsearch bulk indexing semantics and partial-failure handling
        - Avoid unbounded memory growth (streams, batching, large file handling)
        - MCP server compatibility: if indexed document shape, index naming, or query semantics change, validate assumptions against `elastic/semantic-code-search-mcp-server` and explicitly flag any breaking change risk
    - path: 'tests/**/*.ts'
      instructions: |
        Prefer stable tests (no time/locale dependencies, no brittle snapshots).
        Ensure assertions validate behavior, not incidental implementation details.
        For integration tests, keep setup/teardown idempotent and always-run teardown.
    - path: '.github/workflows/**'
      instructions: |
        Review for least-privilege permissions, pinned action SHAs, and caching correctness.
        Ensure CI steps match local scripts (`npm ci`, `npm run lint`, `npm run build`, `npm test`).
    - path: 'scripts/**/*.sh'
      instructions: |
        Prefer strict mode where safe, quote variables, and ensure scripts are idempotent.
        Avoid non-portable bashisms unless the script explicitly requires bash.
    - path: 'Dockerfile'
      instructions: |
        Check for reproducible builds, minimal layers, and avoiding leaking build secrets.
        Prefer non-root runtime where feasible and validate copied artifacts/permissions.

  auto_review:
    enabled: true
    auto_incremental_review: true
    auto_pause_after_reviewed_commits: 5
    drafts: true
    ignore_title_keywords:
      - 'wip'
      - 'do not review'
      - 'skip review'
      - 'draft'

  finishing_touches:
    unit_tests:
      enabled: false
    docstrings:
      enabled: false
    simplify:
      enabled: false