Hi Product Hunt — I'm Michael Curtis, founder of C2Consultants and builder of IntentSharp.

I wanted the intent of an application to be easier for people and coding agents to find, change and check. Too often, the instruction, data shape and rules are scattered across call sites, together with parsing and retry code.

IntentSharp brings those pieces together for .NET applications.

Start with a typed call: give it a value, state what you want, and receive an AIResult<T> with Success, Value, Issues and Attempts. The library checks the answer against your types and rules and can retry with the findings, within the limits you set.

When the wording, shape, rules or examples need to change independently of a release, keep them in versioned YAML or JSON intent files. A saved edit applies on the next call. If a file fails to load, the last good definition stays in force while the application runs.

People and coding agents work with the same definitions. Existing catalog intents can also become AIFunction tools for Microsoft Agent Framework, with each answer checked before the agent receives it.

The checks report what passed. They do not turn a model's answer into a guarantee of correctness. New .NET result types, named code validators and call sites still need a code release.

IntentSharp supports .NET 8, 9 and 10, with OpenAI, Anthropic or any IChatClient.

You can evaluate it free for 30 days on development and test systems. Commercial licences start at $349 per developer per year, with unlimited deployments. Production use requires a paid licence from day one. Personal use and coursework remain free; the library is commercial software, not open source.

Install: dotnet add package IntentSharp

Quick start: https://intentsharp.com/docs/quick-start

Evaluation: https://intentsharp.com/pricing#evaluate

If you're building .NET applications with model calls or coding agents, I'd like to hear which rules you most often need to change — and where the parsing, checking and retry code starts to get in your way.