Show HN: Aqven – typed LLM workflows, checked before a token is spent

Category: ai-ml

Tags: llm-workflows, mcp, static-analysis, agent-tooling, declarative-config

Score: 7.0/10 (Innovation: 7, Technical: 7, Documentation: 8, Utility: 6)

AQVEN is a Python framework and local Studio for building typed LLM workflows as declarative files in a repo, with static checks and simulation before spending tokens. Its distinctive angle is agent-assisted experimentation: a coding agent builds and tests flows via MCP while AQVEN computes verdicts on held-out cases. Strong concept and docs, but alpha-stage, source-available, and requires Python 3.14.

Target audience: AI engineers and backend developers building LLM pipelines

Repository: https://github.com/lastonoga/AQVEN · Python · NOASSERTION · 1 stars

View on Hacker News