You'd reach for this when you need to check LLM inputs for factual consistency or detect semantic bias before processing. It scans text to identify potential alignment issues, which is useful when building systems that need to validate user prompts or filter content through a factual lens. The implementation runs as a remote service on Vercel, so you're making calls over SSE rather than running locally. The source doesn't detail specific operations, but the core use case is clear: pre-flight checks on text heading into language models. Think of it as a gatekeeper that flags when inputs might carry loaded framing or questionable factual claims before they hit your main LLM workflow.
claude mcp add --transport sse io.github.evozim-satori-aligner https://satori-aligner-mcp.vercel.app/api/mcp