Vendor-neutral CLI and MCP server that verifies the token and cost savings AI-coding-agent context-reduction proxies actually deliver, measured against a real labeled corpus.
Enables comparison of responses from multiple LLMs (OpenAI, Anthropic, Gemini) to the same prompt, returning a validated divergence score based on sentence embeddings.
Verifies Ed25519-signed attestation records and builds audit reports. Native TS MCP server exposing verify, ingest, and report tools, thin wrappers over the same aggregator logic the CLI uses.