GEO Phase 3-4: off-site distribution + AI-recommendation measurement loop

[howardpen9]

GEO Phases 3-4, deferred from this session's GEO work (Phases 1-2 — on-site crawlability + definition/comparison pages — are done on branch feat/geo-pricing-sot).

Phase 3 — off-site distribution ("网上得有痕迹")

AI won't recommend a brand that's only described by itself; x402scan alone is thin. Build cross-verifiable third-party presence:

• Directories/registries beyond x402scan Bazaar: awesome-x402 (GitHub list PR), Agent.market, MCP server registries, A2A/agent catalogs; complete the .well-known/x402 discovery doc.
• Genuine third-party traces: Product Hunt launch, a dev tutorial/write-up, Farcaster/X threads others quote, inclusion in an "x402 ecosystem" roundup, Reddit/HN seed. Goal = official + review + directory all say the same thing → AI marks it credible.
• GitHub org presence as a trust anchor (READMEs, examples repo).

Phase 4 — measurement loop

• Fixed prompt set run on ChatGPT / Perplexity / Claude / 豆包: "介绍一下 x402video", "how can an AI agent pay for video generation without an API key", "no-API-key video generation API". Track (1) recognized, (2) recommended on category queries, (3) which pages get cited.
• If a competitor is named instead → targeted comparison page + long-tail "agent-native / pay-per-call video" niche.

Tie into existing GTM cadence (strategy/makereel/social-callout-gtm.md).