version: 1

# Connector keys are the `connection_id` values accepted by
# POST /v1/ingestions. Keep credentials in referenced environment variables.
connectors:
  # The active local connector keeps its source location in the environment.
  harborrag-workspace:
    provider: local
    enabled: true
    environment:
      source_path: LOCAL_SOURCE_PATH
    settings:
      allowed_extensions: [md, markdown, txt, pdf, docx, pptx, csv, xlsx]
      excluded_extensions: []
      include_paths: []
      exclude_paths: []
      include_globs: []
      exclude_globs: []
      include_hidden: false
      follow_symlinks: false
      max_depth: null
      max_file_size_bytes: 104857600
      checksum_mode: stat
      fail_on_error: false

  # Remote connectors are opt-in. Remove the leading "# " from any complete
  # block you need. URLs and source identifiers are resolved from environment
  # variables; credentials remain under secrets.

  # GitHub repository URL is parsed into owner/repository by the provider.
  # engineering-github:
  #   provider: github
  #   enabled: true
  #   environment:
  #     repository_url: GITHUB_REPOSITORY_URL
  #   settings:
  #     branch: main
  #     root_path: docs
  #     allowed_extensions: [md, txt, pdf]
  #     excluded_extensions: []
  #     include_paths: []
  #     exclude_paths: []
  #     include_globs: []
  #     exclude_globs: []
  #     max_file_size_bytes: 104857600
  #     fail_on_error: false
  #     requests_per_minute: 120
  #     request_timeout_seconds: 30
  #     max_retries: 3
  #     backoff_factor: 0.5
  #   secrets:
  #     token_env: GITHUB_TOKEN

  # Confluence URL and space are deployment-specific and stay in the env file.
  # confluence-main:
  #   provider: confluence
  #   enabled: true
  #   environment:
  #     base_url: CONFLUENCE_BASE_URL
  #     space_key: CONFLUENCE_SPACE_KEY
  #   settings:
  #     deployment_type: cloud
  #     content_types: [page]
  #     include_labels: []
  #     exclude_labels: []
  #     include_comments: false
  #     include_attachments: false
  #     max_attachment_size_bytes: 26214400
  #     max_comments: 1000
  #     max_attachments: 1000
  #     max_child_pages: 1000
  #     fail_on_error: false
  #     requests_per_minute: 60
  #     page_size: 25
  #     request_timeout_seconds: 30
  #     max_retries: 3
  #     backoff_factor: 0.5
  #   secrets:
  #     token_env: CONFLUENCE_TOKEN
  #     email_env: CONFLUENCE_EMAIL

  # JIRA URL stays in the environment. Project lists remain typed YAML values.
  # jira-main:
  #   provider: jira
  #   enabled: true
  #   environment:
  #     base_url: JIRA_BASE_URL
  #   settings:
  #     deployment_type: cloud
  #     project_keys: [ENG]
  #     issue_types: []
  #     statuses: []
  #     labels: []
  #     include_comments: true
  #     include_attachments: false
  #     include_changelog: false
  #     include_attachment_text_in_content: true
  #     max_attachment_size_bytes: 26214400
  #     max_comments: 1000
  #     max_attachments: 1000
  #     max_changelog_items: 1000
  #     fail_on_error: false
  #     requests_per_minute: 60
  #     page_size: 50
  #     request_timeout_seconds: 30
  #     max_retries: 3
  #     backoff_factor: 0.5
  #   secrets:
  #     token_env: JIRA_TOKEN
  #     email_env: JIRA_EMAIL

  # SharePoint site URL and Microsoft identity values stay in the environment.
  # knowledge-sharepoint:
  #   provider: sharepoint
  #   enabled: true
  #   environment:
  #     site_url: SHAREPOINT_SITE_URL
  #   settings:
  #     drive_name: Documents
  #     root_path: Shared Documents
  #     allowed_extensions: [md, txt, pdf, docx, pptx, csv, xlsx]
  #     excluded_extensions: []
  #     include_hidden: false
  #     max_file_size_bytes: 104857600
  #     fail_on_error: false
  #     requests_per_minute: 120
  #     page_size: 200
  #     request_timeout_seconds: 60
  #     max_retries: 3
  #     backoff_factor: 0.5
  #   secrets:
  #     tenant_id_env: MICROSOFT_TENANT_ID
  #     client_id_env: MICROSOFT_CLIENT_ID
  #     client_secret_env: MICROSOFT_CLIENT_SECRET