canonical: https://jentic.com/apis/context-link.ai/context-link # Context Link API Jentic publishes the only available OpenAPI specification for Context Link API, keeping it validated and agent-ready. Context Link is a context-aware retrieval API that takes a natural-language query and returns answers grounded in the customer's connected knowledge sources, suitable for RAG-style assistants and chat interfaces. Authentication uses an X-API-Key header and the single GET `/api/v1/context` endpoint accepts a query string and returns the matched context passages. ## For AI agents Retrieve grounded answers from a customer's connected knowledge sources by passing a search query to a single context endpoint authenticated via an X-API-Key header. ## Scope Does not handle source ingestion, embedding configuration, or LLM generation - use for retrieving grounded context passages from connected knowledge sources only. ## Capabilities - Send a natural-language query and receive context passages from the connected knowledge base - Use a single GET `/api/v1/context` endpoint to power retrieval-augmented generation flows - Authenticate per request with an X-API-Key header so each agent or app uses its own scoped key - Pass query strings that include user intent rather than crafting vector embeddings manually - Return ranked passages that an LLM can cite as the source for an answer ## Use cases ### RAG-Powered Customer Support Chatbot Support teams plug Context Link into a customer-facing chatbot to ground responses in their knowledge base, product docs, or help centre. The chatbot sends each user question to GET `/api/v1/context`, then passes the returned passages to an LLM to compose an answer with citations. This avoids the chatbot hallucinating product behaviour and keeps responses tied to vetted source content. Example prompt: GET `/api/v1/context`?query='How do I reset my account password?' with the X-API-Key header, then pass the returned passages to an LLM to draft a cited answer ### Internal Knowledge Assistant Internal IT or HR teams expose a single retrieval endpoint to their employees so policy, process, and onboarding questions resolve against a curated knowledge source. Context Link returns the matched passages so the assistant can answer in plain language and link to the source. Setup is faster than building a vector index because Context Link manages embeddings on the customer's side. Example prompt: Send a user query like 'what is our remote-work policy for engineers' to GET `/api/v1/context` and return the highest-scoring passage with its source link ### AI Agent Retrieval Step via Jentic AI agents that compose multi-step workflows use Context Link as their retrieval step before calling an LLM for synthesis. Through Jentic the agent searches for 'retrieve grounded context', loads the schema, and executes the call with the user query. The X-API-Key is held in your Jentic One instance so the agent never sees the raw key. Example prompt: Search Jentic for 'retrieve grounded context with context link', execute with query='latest API change log', and pass the returned passages to a downstream summarisation step ## Key endpoints | Method | Path | Description | | --- | --- | --- | | GET | `/api/v1/context` | Retrieve grounded context passages for a search query | ## Key resources - **Context** — Retrieval endpoint that returns ranked passages from the connected knowledge sources for a given query ## Why Jentic - **Setup:** Wiring the Context Link API by hand means passing your X-API-Key header and building the context query yourself. Through Jentic you install once, import the Context Link API from the API Directory, store the api_key once, and your agent calls it. - **Permission scoping:** The Context Link API exposes a single read operation that returns grounded context passages, so the agent is limited to that one operation and can do nothing else. Because it only retrieves context from connected sources, there is nothing for the agent to modify or delete. - **Credential handling:** Your Context Link X-API-Key is stored once, encrypted, by your own Jentic One instance and injected into the header at execution time. It never enters the agent's prompt, logs, or context. - **Discovery method:** Agents search Jentic by intent such as 'retrieve grounded context passages', and Jentic returns the single Context Link operation with its query-parameter schema so the agent calls the right endpoint without browsing the reference docs. ## Related APIs - **Pinecone API** — Pinecone is a managed vector database where the developer controls embeddings and indexes directly. - **Exa Search API** — Exa retrieves grounded passages from the open web rather than a customer's connected knowledge sources. - **Tavily Search API** — Tavily provides agent-friendly web search with citations as a single endpoint. - **OpenAI API** — OpenAI generates the final answer from passages returned by Context Link. ## FAQ ### Why is there no official OpenAPI spec for Context Link API? Context Link documents the context endpoint at context-link.ai/docs but does not publish a stable OpenAPI 3 file. Jentic generates and maintains this spec so that AI agents and developers can call Context Link API via structured tooling. It is validated against the live API and kept up to date. Get started with Jentic One, the self-hosted execution layer. ### What authentication does the Context Link API use? Each request must include an X-API-Key header issued from the Context Link dashboard. Through Jentic the API key is stored in your Jentic One instance and applied at call time, so the agent does not see the raw value. ### Can I retrieve grounded answers with the Context Link API? Yes. GET `/api/v1/context` with a query parameter returns the most relevant passages from the connected knowledge sources, which an LLM can then cite when composing an answer. ### What are the rate limits for the Context Link API? Context Link applies per-API-key rate limits sized to the customer's plan; quotas and current usage are visible in the Context Link dashboard. Through Jentic, retries with exponential backoff are handled at the SDK level so transient throttling does not surface to agent code. ### How do I run a retrieval through the Context Link API via Jentic? Run `jentic.search('retrieve grounded context')`, load the operation, and execute with the user's question as the query. Jentic resolves this to GET `/api/v1/context` and returns the passages ready to feed into an LLM. ### Does the Context Link API replace a vector database? Functionally it covers the retrieval step that a vector database plus embedding pipeline would provide, but it is a managed API rather than a database; the customer connects sources and Context Link runs the embedding and ranking. Pair it with Pinecone or Weaviate only if more direct vector control is required. ### Can I limit what my agent is allowed to do with the Context Link API? Yes. Because you run Jentic One yourself, your own rules decide which operations and credentials the agent may use, and the Context Link API exposes only a single read operation, GET `/api/v1/context`, that returns grounded context passages. That means the agent is scoped to retrieving passages from your connected knowledge sources and nothing more, with no way to modify or delete anything. Your stored X-API-Key is injected at call time under your control, so you decide exactly which agent may make that one retrieval call.