DOCSEARCHbyAlgolia
Submit Docs

Kong OpenAPI Specifications

website·Indexed 42 days ago

https://developer.konghq.com

Technical API reference and OpenAPI specifications for Kong Gateway (including Enterprise Edition and Dev Portal) and Kong Konnect. Covers a wide range of services including AI Gateway, AI Catalog, Control Planes, Identity Management, Mesh Manager, and analytics dashboards.

Tokens
952.2K
Snippets
4.1K
Records
7.2K
Agent score
70%

What's inside Kong

  1. Overview of Kong Consumer Groups

    Consumer Groups enable the organization and categorization of Consumers (users or applications) within a Kong API ecosystem. They allow you to scope plugins to specific groups, creating a single plugin instance for the entire group rather than managing individual plugin instances for every consumer. This is particularly useful for managing permissions, roles, resource quotas, and rate limiting across different tiers of users (e.g., Gold vs. Silver tiers).

  2. Overview of MCP servers in Catalog

    An MCP server in Catalog is an interface that represents a Model Context Protocol server your organization builds or consumes. It serves as a centralized inventory to track what MCP servers exist, their capabilities, and how to connect to them, preventing them from being scattered across local config files or repositories.

    Key characteristics:

    • AI Gateway Integration: MCP servers can be linked to AI Gateway 2.0. This indicates that AI Gateway is protecting and proxying that server. Servers not linked are still tracked in the inventory and can be identified as candidates for AI Gateway governance.
    • Snapshot Linking: Linking to AI Gateway is a snapshot; changes in AI Gateway are not automatically synced to the Catalog record.

  3. Overview of Kong Gateway health checks and circuit breakers

    Kong Gateway uses a ring balancer to distribute traffic among Targets. It manages Target availability using two types of health checks:

    1. Active Health Checks: Periodically sends HTTP/HTTPS requests to a specific endpoint on the Target. These can both disable and re-enable Targets automatically based on the response.
    2. Passive Health Checks (Circuit Breakers): Analyzes real-time proxied traffic. These can disable unhealthy Targets but cannot re-enable them automatically.

    Developers can combine both: use passive checks to detect failures and active checks to automatically restore the Target once it recovers. Health checks are disabled by default.

  4. Overview of Kong Gateway Traditional Router

    The traditional router is Kong Gateway's original routing configuration format. It evaluates a collection of Routes against incoming requests until a match is found. Routing based on this JSON configuration is available when the router_flavor is set to traditional_compat or expressions in kong.conf.

  5. Overview of Embedding-based Similarity Matching in AI Gateway

    AI Gateway uses vector embeddings to represent text as points in high-dimensional space. By calculating the distance between these vectors (using metrics like cosine similarity or Euclidean distance), the Gateway can compare the meaning of prompts rather than exact word matches. This powers semantic search for intelligent caching, retrieval-augmented generation (RAG), classification, and anomaly detection.

  6. Overview of AI Consumer Groups

    An AI Consumer Group is an entity in the AI Gateway that represents a collection of AI Consumers. It is used to apply shared AI Policies (such as rate limiting, prompt validation, and PII detection) and access controls to multiple consumers simultaneously, eliminating the need to manage policies for each individual consumer. This allows for scalable AI governance, such as creating subscription tiers (e.g., Bronze, Gold, Enterprise) with different token quotas and cost budgets.

  7. Overview of Konnect Catalog

    Konnect Catalog is a platform-level centralized repository used to create, describe, and govern the interfaces an organization publishes and consumes. It prevents silos by providing a single inventory for APIs, AI Models, MCP servers, and Agents, allowing teams to discover existing resources before building duplicates and enabling security teams to track data providers.

  8. Overview of Kong Gateway Data Plane resilience

    Data Plane resilience allows new Data Plane nodes to be provisioned during a Control Plane (CP) outage. It works by designating backup nodes that have read/write access to a data store (S3-compatible storage). These backup nodes automatically push valid Kong Gateway configurations to the store. When a new Data Plane node is created during a CP outage, it pulls the latest configuration from the store, configures itself, and begins proxying requests.

    Key Constraints:

    • This mechanism only works after a Data Plane node has successfully started. If the node fails to start (e.g., due to unresolved vault references for cluster_cert, cluster_cert_key, or cluster_ca_cert), the fallback mechanism will not engage.
    • The storage volume is only accessed during the creation of the Data Plane node; configurations are not pulled from storage after the node is created.
    • Only backup nodes should have write access to the storage volume.

  9. Overview of Kong AI Gateway

    Kong AI Gateway is a high-performance control plane designed to secure, govern, and observe AI-native systems. It provides a unified layer for managing three primary traffic types: LLM (Large Language Model) calls, MCP (Model Context Protocol) for exposing structured context/tools, and A2A (Agent2Agent) for coordinating AI agents. Key capabilities include routing and load balancing across AI providers, streaming, authentication, ACLs, and usage analytics for tokens, errors, and latency.

  10. Overview of Konnect AI Catalog API Resource Groups

    v1.md

    The Konnect AI Catalog API is organized into the following functional areas:

    • Catalog AI Models: Manage catalogable proxies for AI Gateway models.
    • Catalog AI Model Versions: Manage specific versions of an AI Model.
    • Catalog AI Model Specs: Manage the OpenAPI Specification (OAS) for an AI Model version.
    • Catalog AI Model Implementations: Link a Catalog AI Model to a physical AI Gateway model.
    • Catalog MCPs: Manage catalogable proxies for Model Context Protocol (MCP) servers.
    • Catalog MCP Versions: Manage versions of an MCP.
    • Catalog MCP Implementations: Link a Catalog MCP to an AI Gateway MCP.

  11. Overview of Insomnia's local-first and open-source capabilities

    Insomnia is an open-source (Apache-2.0) desktop client for designing, testing, and documenting APIs. Unlike many competitors, it supports a local-first workflow, meaning users can operate without logging in or relying on cloud workspaces. Key local capabilities include:

    • Local Git-driven storage: Store API definitions, collections, and tests in your own Git repositories; data moves directly between the machine and Git.
    • Local Environment Vault: Encrypt environment values locally without requiring cloud storage.
    • Private-by-default projects: Data remains local or in Git until the user explicitly chooses to sync.
    • Unlimited local usage: No quotas or limits on running requests and collections locally.

  12. Overview of GitLab integration for Catalog Classic

    The GitLab integration allows you to associate a Catalog Classic service with one or more GitLab projects. Once linked, the Konnect UI displays a Project Summary containing data from the GitLab API, including the number of open issues, open merge requests, contributors, languages, and latest releases. Multiple instances of this integration can be configured, including multiple connections to the same provider.