io.github.crunchtools/gemini

Query Gemini with a prompt. Supports Google Search grounding. Args: prompt: The prompt to send to Gemini. model: Model to use ('pro' or 'flash'). use_google_search: Ground response with Google Search results. system_instruction: Optional system instruction to guide the response.

Brainstorm ideas on a topic using Gemini. Args: topic: The topic to brainstorm about. context: Additional context or constraints. num_ideas: Number of ideas to generate. model: Model to use.

Analyze code with Gemini. Args: code: The code to analyze. language: Programming language (auto-detected if not specified). focus: Analysis focus (general, security, performance, bugs). model: Model to use.

Analyze text with Gemini. Args: text: The text to analyze. analysis_type: Type of analysis (general, sentiment, tone, summary). model: Model to use.

Summarize content with Gemini. Args: content: The content to summarize. format: Output format (paragraph, bullets, outline). length: Summary length (brief, moderate, detailed). model: Model to use.

Generate an image from a text prompt using Gemini's native image model. Args: prompt: Description of the image to generate. aspect_ratio: Aspect ratio (1:1, 2:3, 3:2, 3:4, 4:3, 9:16, 16:9, etc). image_size: Output size (1K, 2K, 4K). style: Optional style modifier (photorealistic, cartoon, watercolor, etc). use_google_search: Ground with Google Search results.

Generate/edit an image using a local image file as input. Upload a local image and give Gemini instructions for how to modify it. Use cases: add watermarks, change styles, composite images, etc. Args: prompt: Instructions for what to do with the input image. file_path: Absolute path to a local image file. _aspect_ratio: Desired aspect ratio for the output (currently unused).

Help craft an effective image generation prompt. Args: description: What you want the image to show. style: Desired art style (photorealistic, cartoon, watercolor, etc). mood: Desired mood or atmosphere. model: Model to use for prompt crafting.

Generate images using Google Imagen 4 models. Models (sorted by price): - imagen-4.0-fast-generate-001 ($0.02/image, fastest) - imagen-4.0-generate-001 ($0.04/image, standard) - imagen-4.0-ultra-generate-001 ($0.06/image, highest quality) Args: prompt: Description of the image to generate. model: Imagen model to use. number_of_images: Number of images to generate (1-4). aspect_ratio: Aspect ratio (1:1, 3:4, 4:3, 9:16, 16:9).

Start a multi-turn image editing session. Optionally provide an existing image file as the starting point. Without file_path, Gemini generates a new image from the prompt. Use gemini_continue_image_edit_tool to make further edits. Args: prompt: Initial prompt (describe image or editing instructions). file_path: Optional absolute path to a local image to start from. use_google_search: Ground with Google Search results.

Continue editing in an active image session. Args: session_id: The session ID from gemini_start_image_edit. prompt: Editing instructions (e.g., "make the sky bluer").

End an image editing session and free resources. Args: session_id: The session ID to end.

List all active image editing sessions.

Analyze and describe a local image file. Args: image_path: Absolute path to the image file. query: Question or instruction about the image. model: Model to use ('pro' or 'flash').

Search the web using Gemini with Google Search grounding. Args: query: The search query. model: Model to use.

Analyze a document (PDF, DOCX, TXT, etc.). Args: file_path: Absolute path to the document. question: Question or instruction about the document. model: Model to use.

Summarize a PDF document. Args: file_path: Absolute path to the PDF. style: Summary style (concise, detailed, executive). model: Model to use.

Extract tables from a document. Args: file_path: Absolute path to the document. output_format: Output format (markdown, csv, json). model: Model to use.

Analyze one or more URLs. Args: urls: List of URLs to analyze (1-20). question: Question or instruction about the URL content. use_google_search: Use Google Search grounding. model: Model to use.

Compare two URLs. Args: url1: First URL. url2: Second URL. aspect: What to compare (content, design, sentiment, features). model: Model to use.

Extract specific data from a URL. Args: url: URL to extract data from. data_type: Type of data (text, links, images, structured). custom_fields: Comma-separated list of custom fields to extract. model: Model to use.

Generate a video using Veo. Returns an operation ID to poll. Args: prompt: Description of the video to generate. _aspect_ratio: Video aspect ratio (currently unused). negative_prompt: What to avoid in the video.

Check the status of a video generation operation. Args: operation_name: The operation name from gemini_generate_video.

Analyze a YouTube video. Args: url: YouTube video URL. question: Question or instruction about the video. model: Model to use.

Summarize a YouTube video. Args: url: YouTube video URL. style: Summary style (concise, detailed, bullet_points). model: Model to use.

Convert text to speech. Args: text: Text to convert to speech. voice: Voice name (Zephyr, Puck, Charon, Kore, Fenrir, Leda, Orus, Aoede).

Generate multi-voice dialogue audio. Args: text: Dialogue text with speaker labels. voice1: First voice name. voice2: Second voice name.

List available voices for text-to-speech.

Start a deep research task using a specialized Gemini model. Performs multi-step web research to answer complex questions. Args: query: The research question or topic.

Check the status of a deep research operation. Args: research_id: The research ID from gemini_deep_research.

Ask a follow-up question about completed research. Args: research_id: The research ID from a previous research operation. question: Follow-up question.

Create a content cache for repeated queries. Upload content once and query it multiple times without re-sending. Args: file_path: Path to a file to cache. content: Text content to cache (alternative to file_path). display_name: Human-readable name for the cache. system_instruction: System instruction for queries against this cache. ttl_minutes: Time-to-live in minutes (default 60). model: Model to use.

Query content in a cache. Args: cache_name: The cache name from gemini_create_cache. question: Question to ask about the cached content.

List all active content caches.

Delete a content cache. Args: cache_name: The cache name to delete.

Get structured JSON output from Gemini. Args: prompt: The prompt to generate structured output for. schema: Optional JSON schema to constrain the output. use_google_search: Ground response with Google Search. model: Model to use.

Extract structured data from text. Args: text: Text to extract data from. extract_type: Type (entities, dates, numbers, contacts, custom). custom_fields: Comma-separated list of custom fields to extract. model: Model to use.

Count the number of tokens in content. Args: content: The text content to count tokens for. model: Model to count tokens for.

Execute code using Gemini's built-in code execution. Gemini writes and runs Python code to answer questions, perform calculations, or process data. Args: prompt: Description of what code to write and run. data: Optional data to process. model: Model to use.