sovsparrow

6 ms·

1. 1.▲
   by sovsparrow 2mo ago

   how is it different from microsoft's agent-governance-toolkit?
2. 2.▲
   Show HN: Parquity – find and reproduce Parquet interoperability failures(github.com)

   3 points by sovsparrow 2mo ago|0 comments
3. 3.▲
   by sovsparrow 9mo ago

   i ran into the same “looks legit but wrong” failure mode building modular repos with llms! So i built belgi (released yesterday). thesis: treat llm output as untrusted; acceptance depends on locked spec + executed evidence + invariants (def
4. 4.▲
   by sovsparrow 9mo ago

   Quick links: protocol repo (kernel + canonical specs): https://github.com/belgi-protocol/belgi canonicals/spec: https://github.com/belgi-protocol/belgi/blob/main/CANONICALS...
5. 5.▲
   Show HN: Belgi – deterministic acceptance pipeline for LLM outputs(github.com)

   3 points by sovsparrow 9mo ago|1 comments