Hi everyone,
like almost everyone in software development, we’ve been playing with using AI to help with our work. Most notably we’ve been spending a lot of time in Claude Code.
We’re excited to announce the release of OpenCage Agent Skills to make it simple for Claude and other agentic AI systems to quickly learn about our geocoding API and geosearch service.
The geocoding Skill includes general background about the API and references for installing and using SDKs for various popular programming languages (Python, Node.js, Ruby, Java, etc) as well as our command line geocoding utility. The geosearch Skill trains the AI on how to configure and install our location autosuggest widget.
In both cases the goal is to make it simpler for agents (and software developers) to understand our services and quickly implement them. The new skills are hosted on GitHub and we welcome any pull requests you may have.
The future of AI and geocoding?
We are now frequently getitng asked about whether AI will replace geocoding APIs. So much so that we recently added a “What about AI?” section to our Geocoding Buyer’s Guide. In short, geocoding is a deterministic process where there is a clear correct answer. This is very different than the probabilistic nature of Large Language Models (LLMs). So the correct model is almost certainly for an AI agent to call a geocoding API to get the correct answer. This is also significantly cheaper and faster. And with out new Agent Skills that sort of combined architecture is now easier than ever.
Try It Now
If agentic coding is your thing please take our new Agent Skills for a spin (see the install.md file to get started (or just ask Claude to install it), and let us know how it works out. We welcome your feedback.
Happy agentic geocoding,
Note - this post was written by a human, not an agent!