TrueFoundry · Capability

TrueFoundry LLM Operations

Workflow capability for AI/LLM operations via TrueFoundry's AI Gateway. Combines chat completions, embeddings, image generation, content moderation, and document reranking into a unified AI operations workflow. Used by AI engineers, data scientists, and application developers building LLM-powered applications on TrueFoundry's enterprise platform.

Run with Naftiko TrueFoundryAI GatewayLLMChat CompletionsEmbeddingsImage GenerationContent ModerationRerankingMLOps

What You Can Do

POST
Create chat completion — Generate chat completion from any supported model
/v1/chat/completions
POST
Create embeddings — Generate vector embeddings
/v1/embeddings
POST
Generate images — Generate images from text prompts
/v1/images/generations
POST
Create moderation — Check content for policy violations
/v1/moderations
POST
Rerank documents — Rerank documents by query relevance
/v1/rerank
GET
List models — List all available models
/v1/models
POST
Create batch — Create a batch job
/v1/batches
GET
List files — List uploaded files
/v1/files

MCP Tools

chat-completion

Generate a chat completion using any model available in TrueFoundry AI Gateway (GPT-4o, Claude, Gemini, Mistral, etc.). Supports streaming, tool calling, and all OpenAI-compatible parameters.

create-embeddings

Generate vector embeddings for text. Used for semantic search, RAG pipelines, and similarity comparisons across 1000+ embedding models.

read-only idempotent
generate-images

Generate images from text prompts using AI image models

moderate-content

Check text content for policy violations

read-only idempotent
rerank-documents

Rerank a list of documents by their relevance to a query. Used in RAG pipelines to improve retrieval quality.

read-only idempotent
list-models

List all LLM models available in TrueFoundry AI Gateway

read-only
create-batch

Create a batch job for async processing of multiple LLM requests

list-files

List files uploaded to TrueFoundry

read-only

APIs Used

tfy-gateway

Capability Spec

llm-operations.yaml Raw ↑ 
naftiko: "1.0.0-alpha1"

info:
  label: "TrueFoundry LLM Operations"
  description: >-
    Workflow capability for AI/LLM operations via TrueFoundry's AI Gateway.
    Combines chat completions, embeddings, image generation, content moderation,
    and document reranking into a unified AI operations workflow. Used by AI
    engineers, data scientists, and application developers building LLM-powered
    applications on TrueFoundry's enterprise platform.
  tags:
    - TrueFoundry
    - AI Gateway
    - LLM
    - Chat Completions
    - Embeddings
    - Image Generation
    - Content Moderation
    - Reranking
    - MLOps
  created: "2026-05-03"
  modified: "2026-05-03"

binds:
  - namespace: env
    keys:
      TRUEFOUNDRY_API_KEY: TRUEFOUNDRY_API_KEY

capability:
  consumes:
    - import: tfy-gateway
      location: ./shared/ai-gateway.yaml

  exposes:
    - type: rest
      port: 8080
      namespace: truefoundry-llm-api
      description: "Unified REST API for LLM operations via TrueFoundry AI Gateway."
      resources:
        - path: /v1/chat/completions
          name: chat-completions
          description: "Chat completion for LLM conversations"
          operations:
            - method: POST
              name: create-chat-completion
              description: "Generate chat completion from any supported model"
              call: "tfy-gateway.create-chat-completion"
              with:
                model: "rest.model"
                messages: "rest.messages"
                temperature: "rest.temperature"
                max_tokens: "rest.max_tokens"
              outputParameters:
                - type: object
                  mapping: "$."
        - path: /v1/embeddings
          name: embeddings
          description: "Text embedding generation"
          operations:
            - method: POST
              name: create-embeddings
              description: "Generate vector embeddings"
              call: "tfy-gateway.create-embeddings"
              with:
                model: "rest.model"
                input: "rest.input"
              outputParameters:
                - type: object
                  mapping: "$."
        - path: /v1/images/generations
          name: image-generations
          description: "AI image generation"
          operations:
            - method: POST
              name: generate-images
              description: "Generate images from text prompts"
              call: "tfy-gateway.generate-images"
              with:
                model: "rest.model"
                prompt: "rest.prompt"
              outputParameters:
                - type: object
                  mapping: "$."
        - path: /v1/moderations
          name: moderations
          description: "Content moderation"
          operations:
            - method: POST
              name: create-moderation
              description: "Check content for policy violations"
              call: "tfy-gateway.create-moderation"
              with:
                input: "rest.input"
              outputParameters:
                - type: object
                  mapping: "$."
        - path: /v1/rerank
          name: rerank
          description: "Document reranking"
          operations:
            - method: POST
              name: rerank-documents
              description: "Rerank documents by query relevance"
              call: "tfy-gateway.rerank-documents"
              with:
                model: "rest.model"
                query: "rest.query"
                documents: "rest.documents"
              outputParameters:
                - type: object
                  mapping: "$."
        - path: /v1/models
          name: models
          description: "Available model catalog"
          operations:
            - method: GET
              name: list-models
              description: "List all available models"
              call: "tfy-gateway.list-models"
              outputParameters:
                - type: object
                  mapping: "$."
        - path: /v1/batches
          name: batches
          description: "Batch request management"
          operations:
            - method: POST
              name: create-batch
              description: "Create a batch job"
              call: "tfy-gateway.create-batch"
              with:
                input_file_id: "rest.input_file_id"
                endpoint: "rest.endpoint"
              outputParameters:
                - type: object
                  mapping: "$."
        - path: /v1/files
          name: files
          description: "File management"
          operations:
            - method: GET
              name: list-files
              description: "List uploaded files"
              call: "tfy-gateway.list-files"
              outputParameters:
                - type: object
                  mapping: "$."

    - type: mcp
      port: 9090
      namespace: truefoundry-llm-mcp
      transport: http
      description: "MCP server for AI-assisted LLM operations via TrueFoundry."
      tools:
        - name: chat-completion
          description: >-
            Generate a chat completion using any model available in TrueFoundry
            AI Gateway (GPT-4o, Claude, Gemini, Mistral, etc.). Supports
            streaming, tool calling, and all OpenAI-compatible parameters.
          hints:
            readOnly: false
            openWorld: true
          call: "tfy-gateway.create-chat-completion"
          with:
            model: "tools.model"
            messages: "tools.messages"
            temperature: "tools.temperature"
            max_tokens: "tools.max_tokens"
          outputParameters:
            - type: object
              mapping: "$."
        - name: create-embeddings
          description: >-
            Generate vector embeddings for text. Used for semantic search,
            RAG pipelines, and similarity comparisons across 1000+ embedding models.
          hints:
            readOnly: true
            idempotent: true
          call: "tfy-gateway.create-embeddings"
          with:
            model: "tools.model"
            input: "tools.input"
          outputParameters:
            - type: object
              mapping: "$."
        - name: generate-images
          description: "Generate images from text prompts using AI image models"
          hints:
            readOnly: false
          call: "tfy-gateway.generate-images"
          with:
            model: "tools.model"
            prompt: "tools.prompt"
            n: "tools.n"
            size: "tools.size"
          outputParameters:
            - type: object
              mapping: "$."
        - name: moderate-content
          description: "Check text content for policy violations"
          hints:
            readOnly: true
            idempotent: true
          call: "tfy-gateway.create-moderation"
          with:
            input: "tools.input"
          outputParameters:
            - type: object
              mapping: "$."
        - name: rerank-documents
          description: >-
            Rerank a list of documents by their relevance to a query. Used
            in RAG pipelines to improve retrieval quality.
          hints:
            readOnly: true
            idempotent: true
          call: "tfy-gateway.rerank-documents"
          with:
            model: "tools.model"
            query: "tools.query"
            documents: "tools.documents"
            top_n: "tools.top_n"
          outputParameters:
            - type: object
              mapping: "$."
        - name: list-models
          description: "List all LLM models available in TrueFoundry AI Gateway"
          hints:
            readOnly: true
          call: "tfy-gateway.list-models"
          outputParameters:
            - type: object
              mapping: "$."
        - name: create-batch
          description: "Create a batch job for async processing of multiple LLM requests"
          hints:
            readOnly: false
          call: "tfy-gateway.create-batch"
          with:
            input_file_id: "tools.input_file_id"
            endpoint: "tools.endpoint"
            completion_window: "tools.completion_window"
          outputParameters:
            - type: object
              mapping: "$."
        - name: list-files
          description: "List files uploaded to TrueFoundry"
          hints:
            readOnly: true
          call: "tfy-gateway.list-files"
          outputParameters:
            - type: object
              mapping: "$."