Write audit_protocol(spec) that takes a protocol spec (dict) and
returns a dict with two fields:
score: integer 0-100, higher means more likely to WINverdict: string, one of:"will-win"if score >= 75"maybe"if score >= 50"niche"if score >= 25"dead-on-arrival"if score < 25
Score the spec on these four signals (each worth 25 points when True):
open_spec: 25 points (anyone can implement)multi_client: 25 points (more than one host adopts it)low_publisher_bar: 25 points (one file, no marketplace review)primitive_not_marketplace: 25 points (developer wires it up, not end-user browse)
Two protocols run. Expected output:
MCP: {'score': 100, 'verdict': 'will-win'}
OpenAI Plugins: {'score': 0, 'verdict': 'dead-on-arrival'}