I’m building an open-source decision layer above AI agents [P]

In three linesSpice is an open-source decision layer above AI agents. It makes the decision process explicit (observations, options, justifications, trade-offs) before execution, rather than treating agents as black boxes. Compatible with Claude Code, Codex, and other executors.

Summary generated by Claude — human-verified