Agenci brings CI/CD to AI agents.Run functional tests,security policy checks and r egression diffs against your agent before every change ships - catch prompt injection vulnerabilities,tool-authorization violations,cost blowups and quality regressions automatically.Works with OpenAI,Anthropic,LangChain,Crew AI,AutoGen or plain HTTP/Python agents.Free and open source.
Hello ProductHunt!Maker here.Built Agenci after I saw that there is no real testing infrastructure for AI agents,even though their behavior can regress just like code can (less accurate,more expensive,less secure) with zero warning.
Tested it against a small local model and it caught a real prompt-injection vulnerability - told the model to ignore its instructions and it complied instantly.
Free,open source,runs locally.Would love feedback,especially from anyone building or shipping agents in production.