Pexafy

Pexafy provides semantic search over millions of free photos from stock libraries including Unsplash, Pexels and Pixabay, returning real photographic works along with license information and credit text when needs are described in natural language.

Getting started with Pexafy requires only a free API Key (no card binding needed) and a key from any model provider; the official example uses Claude Haiku 4.5, but it emphasizes that any provider works.

Pexafy provides an MCP server (https://mcp.pexafy.com/mcp), using the API Key as a Bearer request header, and can be connected in the same way by Claude Desktop, Cline and other MCP-supporting hosts.

Pexafy's native integration in LangChain/LangGraph (Python) installs langchain-pexafy, langchain and langchain-anthropic via pip, and the core code calls create_agent and passes in PexafyToolkit().get_tools(); the toolkit contains three tools: search by description, find similar images, and get a single photo.

In the TypeScript ecosystem, Pexafy provides pexafy-ai-sdk, used together with ai, zod and @ai-sdk/anthropic, and can be integrated by passing pexafyTools() into generateText; the example limits multi-step call rounds via stopWhen: stepCountIs(5).

The Pexafy tools return structured records containing fields such as photo_id, alt_text, url, width, height, orientation, photographer, source, license and credit.

Pexafy's free tier requires a credit line next to each published image, and automated workflows using this service should write this constraint into the agent instructions as a rule.

alt_text is the value of the alt attribute in HTML, used to provide a text description when the image cannot be displayed, and is also the basis for screen readers to read out image content for visually impaired users, as well as an important signal for search engines to understand image semantics; the alt_text returned by Pexafy is written manually by the stock library's photographers or platform.

Pexafy's official documentation lists tested configurations for the OpenAI Agents SDK, Claude Agent SDK, Google ADK, CrewAI, LlamaIndex, Mastra and Strands Agents.

In scenarios requiring similar image recommendations, search_photos_by_image is more efficient than text search; when semantic search results are unsatisfactory, the agent should rewrite the query and retry or call search_photos_by_image to find similar assets using an existing image as reference.

For scenarios generating content in bulk, it is recommended to add an automated check step before the final output to scan whether all images on the page are accompanied by credits, in order to avoid copyright risk, so that the agent can run the image-sourcing step unattended.