A large language model (LLM) is an AI model trained on massive datasets to generate human-like responses. Deep-learning techniques such as transformers help it understand contextual relationships, syntax, and semantics, enabling content generation, translation, and conversation. section 3.10
In software development, an LLM can:
- Power AI coding assistants that analyze code, generate functions, suggest completions, and propose bug fixes. section 3.10
- Apply natural-language rules to OpenAPI documents and report linting violations that traditional rules may not easily express. section 3.11
- Reduce the cost and time of checking changed files, while leaving developers responsible for testing and reviewing generated results. section 3.10 section 3.11
OpenAPI monetary-field linting
Upload an OpenAPI document to Claude and ask it to find response properties whose names imply monetary amounts. It identifies cost_price, selling_price, and vat_amount, then reports that each is incorrectly defined as type: string rather than type: number. The same natural-language rules can run through Optic’s LintGPT in a CI/CD pipeline. section 3.11.1 section 3.11.2
Lint the OpenAPI document based on the following rule: - If a property’s name implies it is a monetary amount, it must use number format. Show me the linting rule violations. An example of a violation can be: “The property ‘cost_price’ implies a monetary amount but is defined with 'type: string'. It should be defined with 'type: number' to accurately represent a monetary value."
Shows the natural-language linting prompt used to check monetary property names and required number formats. section 3.11.1
