///
Fabric provides a robust set of advanced features designed to augment human capabilities using AI. These functionalities extend beyond basic [pattern execution](/wiki/danielmiessler/Fabric/patterns-pr
278 views
~278 views from guests
Guest views are estimated from total page views. These include anonymous visitors and users who weren't logged in when they viewed the page.
Fabric provides a robust set of advanced features designed to augment human capabilities using AI. These functionalities extend beyond basic pattern execution, offering sophisticated control over AI interactions, from web integration to multimedia processing and fine-tuned model reasoning.
Fabric integrates web search capabilities directly into its AI interactions, allowing supported models (Anthropic, OpenAI, and Gemini) to access and incorporate real-time information from the internet. This is powered by Jina AI, which also provides dedicated web scraping capabilities.
To enable web search for a chat request:
--search: Activates the web search tool for the AI model, allowing it to perform queries as needed.--search-location=: Specifies a geographical location for search results, influencing the relevance and localization of information (e.g., --search-location='America/Los_Angeles').For more general web content extraction, see the --scrape_url and --scrape_question flags in the CLI Usage Guide.
Fabric supports AI-powered image generation, allowing users to create images with precise control over their attributes. This feature, primarily leveraging OpenAI models, provides detailed customization options.
To generate and save images:
--image-file=: Specifies the file path where the generated image will be saved (e.g., output.png). This flag is mandatory for image generation.--image-size=: Sets the dimensions of the generated image. Available options include 1024x1024, 1536x1024, 1024x1536, or auto (default: auto).--image-quality=: Controls the quality level of the image. Options are low, medium, high, or auto (default: auto).--image-compression=: Adjusts the compression level for generated JPEG/WebP images, ranging from 0 to 100 (default: not set). This flag is only applicable to JPEG and WebP formats.--image-background=: Defines the background type for the image. Options are opaque or transparent (default: opaque). Transparency is only supported for PNG and WebP formats.Fabric provides robust Text-to-Speech (TTS) capabilities, particularly with Gemini models, allowing you to convert AI-generated text into spoken audio.
--voice=: Selects a specific voice for the TTS output (e.g., Kore, Charon, Puck). The default voice is Kore.--list-gemini-voices: Displays a comprehensive list of all available Gemini TTS voices.-o, --output= flag (e.g., -o output.wav). If an audio output file is specified without a TTS model, or a TTS model is chosen without an audio output file, Fabric will prompt you with an error.Fabric offers Speech-to-Text (STT) functionality, primarily through OpenAI models, enabling transcription of audio and video files into text.
--transcribe-file=: Specifies the path to the audio or video file that needs to be transcribed.--transcribe-model=: Selects the AI model to be used for the transcription task (e.g., whisper-1). This is separate from the model used for chat.--split-media-file: Automatically splits audio/video files larger than 25MB into smaller chunks using ffmpeg before transcription, to comply with API limits.--list-transcription-models: Provides a list of available models that can be used for transcription.Fabric allows for fine-grained control over the AI model's reasoning process, enabling users to adjust the depth of thought and manage how internal "thinking" blocks are presented in the output.
--thinking=: Sets the reasoning or "thinking" level of the model. Options include off, low, medium, high, or a specific numeric token value (for Anthropic and Google Gemini models) to allocate for internal thought processes. Higher levels of thinking can lead to more elaborate and coherent responses, albeit potentially with increased processing time or cost.--suppress-think: Hides content enclosed within designated "thinking" tags in the model's output. This is useful for obtaining only the final, distilled answer without the intermediate thought process.--think-start-tag=: Defines a custom starting tag for thinking sections (default: <think>).--think-end-tag=: Defines a custom ending tag for thinking sections (default: </think>).These flags provide powerful tools for enhancing and refining your interactions with AI models through Fabric. For a complete overview of all command-line options, refer to the CLI Usage Guide.