mirror of
https://github.com/katanemo/plano.git
synced 2026-06-23 15:38:07 +02:00
16 lines
4.5 KiB
ReStructuredText
16 lines
4.5 KiB
ReStructuredText
### test archfc with plano orchestrator
|
|
POST https://archfc.katanemo.dev/v1/chat/completions HTTP/1.1
|
|
Content-Type: application/json
|
|
model: Plano-Orchestrator
|
|
|
|
{
|
|
"model": "Plano-Orchestrator",
|
|
"messages": [
|
|
{
|
|
"role": "user",
|
|
"content": "You are a helpful assistant that selects the most suitable routes based on user intent.\nYou are provided with a list of available routes enclosed within <routes></routes> XML tags:\n<routes>\n{\"name\": \"Harvey\", \"description\": \"Harvey is a professional-class AI platform built for industry leaders. It offers a suite of tools to augment various work processes, including research, document management, and task delegation. The platform emphasizes security and integrates domain-specific models for complex professional work. Harvey is used by leading law firms and Fortune 500 companies.\n\nCapabilities: \n * Delegates complex tasks via natural language to a domain-specific personal assistant.\n * Provides answers rooted in reliable, cited source material from up to 50 documents.\n * Supports drafting and revising complex long-form content.\n * Offers pre-crafted prompts to streamline workflows.\n * Operates in 50+ languages, countries, and legal systems.\n * Integrates with Microsoft Word to enhance drafting capabilities.\", \"parameters\": {\"type\": \"object\", \"properties\": {}, \"required\": []}}\n{\"name\": \"Mirtilla\", \"description\": \"Mirtilla is an AI-powered platform designed to revolutionize meetings and notes. It offers robust transcription capabilities, including translation and timestamping, allowing for efficient review and analysis of discussions. The platform also provides concise summaries and records of meetings, highlighting essential points and decisions. Users can generate custom requests tailored to provide insights, define focus areas, or create specific outputs. Mirtilla's AI Notes feature allows users to input notes with tags for classification and utilize AI to search through these notes using natural language. Data is fully encrypted with zero-knowledge encryption, ensuring privacy and compliance with GDPR regulations. The platform is hosted on Italian servers.\n\nCapabilities: \n * Meeting Transcription: Transcribes audio/video files, translates to English, and includes timestamping.\n * Meeting Summarization: Generates concise summaries, minutes, and records of meetings.\n * Personalized Requests: Creates custom outputs from transcription content, such as LinkedIn posts or summaries.\n * AI-Powered Notes: Allows note input with tags, AI-powered natural language search, and sharing capabilities.\", \"parameters\": {\"type\": \"object\", \"properties\": {}, \"required\": []}}\n{\"name\": \"ThumbGenie\", \"description\": \"ThumbGenie is the #1 AI thumbnail maker for creating high-converting, personalized YouTube thumbnails automatically. Our powerful thumbnail generator customizes designs to match your channel's unique style, boosting click-through rates while saving you time and money.\n\nCapabilities: \n * Generate multiple YouTube thumbnail designs automatically.\n * Customize designs to match your channel's unique style.\n * Train AI models based on your existing thumbnails for consistent branding.\n * Adjust realism scale to fine-tune thumbnail style.\n * Use a YouTube video URL to add a reference image.\", \"parameters\": {\"type\": \"object\", \"properties\": {}, \"required\": []}}\n</routes>\n\nYou are also given the conversation context enclosed within <conversation></conversation> XML tags:\n<conversation>\n[\n {\n \"role\": \"user\",\n \"content\": \"Hey, how's it going?\"\n },\n {\n \"role\": \"assistant\",\n \"content\": \"Hello! I'm doing well, thank you. How can I assist you today?\"\n },\n {\n \"role\": \"user\",\n \"content\": \"I need to draft a response to a legal motion. It's fairly complex and needs to cite several case precedents.\"\n }\n]\n</conversation>\n\n## Instructions\n1. Analyze the latest user intent from the conversation.\n2. Compare it against the available routes to find which routes can help fulfill the request.\n3. Respond only with the exact route names from <routes>.\n4. If no routes can help or the intent is already fulfilled, return an empty list.\n\n## Response Format\nReturn your answer strictly in JSON as follows:\n{\"route\": [\"route_name_1\", \"route_name_2\", \"...\"]}\nIf no routes are needed, return an empty list for `route`."
|
|
}
|
|
],
|
|
"continue_final_message": false,
|
|
"add_generation_prompt": true
|
|
}
|