Configuration

How to configure config.yaml

You can directly modify it in the TUI under Configuration → General Settings (Recommended), or manually edit config/config.yaml.

💡 The actual config file is located at C:\Users\{username}\AppData\Roaming\mailslide\config\config.yaml

llm_mode: per_plugin
plugin_modules:
  - custom_plugins.my_plugin_module

jobs:
  - name: "My Job"
    account: "[email protected]"
    source: "Inbox"
    destination: "Inbox/processed"
    manual_review_destination: "Inbox/manual_review"
    batch_flush_enabled: true
    limit: 10
    plugins:
      - add_category

Field	Description
name	Job name (can be named arbitrarily)
enable	Whether to enable this job (true/false, default true)
account	Outlook account Email or PST file name
source	Source folder (e.g., Inbox)
destination	Destination folder after processing (optional; if omitted, emails won't be moved and might be processed repeatedly, can be omitted for simple testing)
manual_review_destination	Folder for skipped/failed LLM outcomes to be manually reviewed (optional)
limit	Number of emails to process
llm_mode	LLM call mode (per_plugin default; share_deprecated for legacy mode)
plugin_modules	List of additional plugin module paths (dynamically imported at startup for registering custom plugins)
ui_language	UI language (zh-TW / en-US, default en-US)
batch_flush_enabled	Job-level batch writing toggle (default true; affects some file writing modules)
plugins	List of enabled plugins (optional)

💡 Tip: If you use the move_to_folder plugin to let the LLM decide the destination directory, you can omit destination, and the plugin will handle moving the emails.

LLM Call Modes

• per_plugin (default): Each plugin needing an LLM calls the LLM independently. Suitable for mixing multiple plugins.
• share_deprecated (legacy mode): One LLM call shared across all LLM plugins for a single email. This mode easily causes action_mismatch/skipped due to different action values and is not recommended for new configurations.

llm_mode can be set at:

• Global level: config.llm_mode
• Job level override: job.llm_mode

Backward compatibility: Legacy values shared / shared_legacy can still be executed, but they will be treated as share_deprecated and log a warning.

Configure LLM (Optional)

If you want to use plugins integrated with AI analysis features, it's recommended to edit the configuration in the TUI under Configuration → LLM Settings; you can also manually edit config/llm-config.yaml:

api_base: "http://localhost:11434/v1"
api_key: "your-api-key"
model: "llama3"

Supports OpenAI-compatible API formats such as Ollama and LM Studio.