A growing collection of real-world examples built with Lamia — a framework for writing AI-powered scripts in plain English.
New examples are added continuously. Star this repo to get notified on updates.
Lamia is all about preciseness — getting 100% of the expected results, every time. You write what you want; Lamia handles the LLM calls, validates the output, and returns structured data. No guessing, no loose outputs.
If a Lamia script fails, it usually means one of two things:
- The prompt needs to be adjusted for the task at hand
- The LLM model being used isn't capable enough for that particular output
Each example ships with its own config.yaml that specifies the exact models the example was tested against. Using that config.yaml is the recommended starting point — it removes model variance from the equation so you can focus on understanding the example itself.
If a script is still failing when using the example's own
config.yaml, please open an issue.
| Example | Description |
|---|---|
| prd_implementor | A virtual software engineering team that takes a PRD markdown file and implements it end-to-end — PM, groomer, developer, reviewer, and QA analyst, all orchestrated by a single .lm script |
Have a use case you'd like to see covered with Lamia? Two ways to contribute:
- Request an example — open an issue describing the use case. If accepted and developed, the code will be published here. Note that it will be open sourced under the MIT license.
- Submit an example — open a pull request with your Lamia example. Each example should include its own
README.md,config.yaml, and be self-contained in its own directory.