sondera-ai

Turns one atomic governance intent ("the agent must not <X>") into a validated Cedar forbid policy for the Sondera OSS harness, or into a refusal that names what the engine cannot express. Drives the `sondera` MCP server (`crates/mcp`, run by `cargo run -p…

Authors, validates, and tests Cedar authorization policies and schemas for Sondera AI agent governance. Use when writing .cedar or .cedarschema files, working with agent guardrails, YARA signatures, sensitivity labels, or information flow control policies.