NagentArtefacts
Dev & Technology
Enterprise and White Glove
Company
APIpie AI connects to a Nagent workspace with an API key. Once it is connected, agents can call 13 APIpie AI actions, such as "Create Vector Collection", "Get Detailed Models" and "Get query history". Nothing is enabled on connect: each action is allowed one at a time, and an action with side effects runs or waits for a person according to the agent's level.
Every operation an agent can call against APIpie AI, with input parameters and output schema.
APIPIE_AI_ANONYMIZE_TEXTAnonymize sensitive entities (PII) in text for data privacy and compliance. Use this tool to detect and replace personally identifiable information like names, phone numbers, locations, and other sensitive data with truncated SHA-256 hashes. Returns both the anonymized text and mappings showing what was replaced.
Input parameters
Text containing potentially sensitive information (PII) to be anonymized. The API will detect and replace entities like names, phone numbers, locations, etc. with truncated SHA-256 hashes.
Output
Data from the action execution
Error if any occurred during the execution of the action
Whether or not the action execution was successful or not
APIPIE_AI_CREATE_VECTOR_COLLECTIONCreate a new vector collection (Pinecone-style index and namespace combined) in APIpie. Use this when you need to set up a new vector database for storing embeddings with a specific dimension. The dimension must match the embedding model you'll use (e.g., 1536 for OpenAI text-embedding-ada-002).
Input parameters
Vector dimension size. Common values: 512, 768, 1536, 3072. Must match the embedding model dimension you plan to use.
Name for the vector collection. Must be unique within your account.
Output
Data from the action execution
Error if any occurred during the execution of the action
Whether or not the action execution was successful or not
APIPIE_AI_DELETE_STATETool to delete state settings from APIpie. Without query parameter deletes app-level state; with query parameter deletes specific user state. Use after configuring state to remove unwanted state records or reset configuration.
Input parameters
Optional. If provided, delete state for the specified user. If omitted, delete all state records for this API key/app.
Optional. Enables the ability for centralized API key state management for enterprises.
Output
Data from the action execution
Error if any occurred during the execution of the action
Whether or not the action execution was successful or not
APIPIE_AI_DELETE_VECTORSDelete vectors from a vector collection in APIpie. Use this tool to: - Delete ALL vectors in a collection: set delete_all=True (requires credits) - Delete specific vectors: set delete_all=False and provide a list of vector IDs Note: Deleting by metadata filter is not currently supported - you must specify vector IDs.
Input parameters
List of vector IDs to delete. Required when delete_all=False. Example: \['vec-001', 'vec-002'\].
Reserved for future use. Metadata filter for deletion is not currently supported by the API.
If True, deletes ALL vectors in the collection (requires credits). If False, you must provide 'ids' to specify which vectors to delete.
Name of the vector collection to delete from. Must match an existing collection.
Output
Data from the action execution
Error if any occurred during the execution of the action
Whether or not the action execution was successful or not
APIPIE_AI_GET_DETAILED_MODELSFetch detailed information about available AI models including pricing, capabilities, and specifications. Use when you need comprehensive model data with pricing rates, token limits, modality support, and benchmark scores.
Input parameters
Filter by type of model (llm, vision, embedding, image, voice, moderation, code)
Filter by model name
Filter by provider (e.g., openai, anthropic)
Output
Data from the action execution
Error if any occurred during the execution of the action
Whether or not the action execution was successful or not
APIPIE_AI_GET_QUERY_HISTORYTool to retrieve historic API usage logs including latency, token counts, costs, and source IP. Use after authenticating to analyze past queries for cost management, performance monitoring, or auditing.
Input parameters
Limits the number of returned queries per page.
Offset used for pagination.
Output
Data from the action execution
Error if any occurred during the execution of the action
Whether or not the action execution was successful or not
APIPIE_AI_GET_STATETool to retrieve current state settings including user preferences, memory configuration, and routing settings. Use when you need to check or audit the current configuration for an app or specific user.
Input parameters
Optional. If provided, returns the state for the specified user. If omitted, returns the app-level (master) state for the current API key/app.
Optional. Enables the ability for centralized API key state management for enterprises.
Output
Data from the action execution
Error if any occurred during the execution of the action
Whether or not the action execution was successful or not
APIPIE_AI_LIST_MODELSFetch a list of available AI models from APIPie. Use this tool when you need: - Up-to-date model listings with filtering by type, subtype, or provider - Voice model listings (set voices=true) - Country restriction information (set restrictions=true) Returns models with pricing, latency, availability, and capability information.
Input parameters
Filter by model type (e.g., 'llm' for language models)
Set to true to retrieve only voice models list (use as flag)
Filter by enabled status (1 for enabled, 0 for disabled)
Filter by model subtype (e.g., 'chat')
Filter by provider (e.g., 'openrouter')
Filter by provider and subtype combination (e.g., 'provider=openrouter&subtype=chat')
Set to true to retrieve only country restrictions list (use as flag)
Output
Data from the action execution
Error if any occurred during the execution of the action
Whether or not the action execution was successful or not
APIPIE_AI_LIST_VECTOR_COLLECTIONSTool to retrieve a list of all vector collections under your account. Use when you need to view available collections before performing vector operations like querying, upserting, or deleting vectors.
Output
Data from the action execution
Error if any occurred during the execution of the action
Whether or not the action execution was successful or not
APIPIE_AI_PARSE_DOCUMENTTool to parse document content and metadata using Apache Tika. Extract text and metadata from various document formats (PDF, DOCX, TXT, etc.). Use when you need to extract readable text or metadata from uploaded documents.
Input parameters
Document file to parse.
Extract text content from document. Defaults to true.
Extract metadata from document. Defaults to true.
Output
Data from the action execution
Error if any occurred during the execution of the action
Whether or not the action execution was successful or not
APIPIE_AI_TRANSCRIBE_AUDIOTool to transcribe audio files to text using AI speech-to-text models like Whisper. Use when you need to convert spoken audio into written text. Supports multiple models and output formats.
Input parameters
Audio file to transcribe (max 25MB).
Model to use for transcription. Options: whisper-1 (OpenAI), gpt-4o-transcribe, gpt-4o-mini-transcribe.
Optional text prompt to guide the model and improve transcription quality. Can include specific terminology or context.
Whether to stream the response. Set to true for streaming transcription output.
ISO 639-1 or 639-3 language code of the audio (e.g., 'en', 'es', 'fr'). Helps improve transcription accuracy.
Sampling temperature between 0 and 1. Lower values make output more deterministic.
Response format options for transcription.
Timestamp granularity for transcription. Only works with whisper-1 model when response_format is 'verbose_json'. Options: 'word', 'segment'.
Output
Data from the action execution
Error if any occurred during the execution of the action
Whether or not the action execution was successful or not
APIPIE_AI_UPDATE_STATETool to create or update state settings in APIpie, including configurations, deletions, and feature toggling at app or user levels. Use when you need to manage persistent state for AI completions, memory, routing, or other APIpie features.
Input parameters
Required when enable_user_states=true. Identifies the end-user for user-scoped state.
Keys to remove from existing settings (partial delete). Use POST to perform partial key deletes.
Optional. Enables centralized API key state management for enterprises.
When true, state configuration can only be performed through APIpie GUI and cannot be edited via API or inline CLI.
Settings persisted per user or per app (API key) depending on scope.
Enable command line CLI for state management at the prompt. Use ':help' in prompt to any LLM for details.
If true, scope is per-user (user_id + user). If false, scope is per-app (user_id + app label).
Output
Data from the action execution
Error if any occurred during the execution of the action
Whether or not the action execution was successful or not
APIPIE_AI_UPLOAD_FILEUpload a file to APIPie and retrieve a temporary URL. Use when you need to upload an image file and get a shareable URL. Supports image formats (.png, .jpg, .jpeg, .svg, .gif, .bmp, .tif, .tiff, .webp) with a maximum size limit of 5MB.
Input parameters
File to upload. Supported formats:.png.jpg.jpeg.svg.gif.bmp.tif.tiff.webp. Max size: 5MB.
Output
Data from the action execution
Error if any occurred during the execution of the action
Whether or not the action execution was successful or not
Agents can call 13 APIpie AI actions on Nagent, including "Create Vector Collection", "Get Detailed Models" and "Get query history". Create Vector Collection: Create a new vector collection (Pinecone-style index and namespace combined) in APIpie. Each action is listed on this page with its input parameters and its output.
APIpie AI connects with an API key, under your workspace's own connection. Nothing is enabled on connect: each action is allowed one at a time and can be scoped to the agents that need it.
Create Vector Collection takes 2 required inputs: dimension and collection_name. It returns data, error and successful.