Great point @alexshev — downloads ARE a vanity metric. The real signal is: did the agent connect, complete a workflow, and self-diagnose failures?
We are building toward exactly that. Currently exposed:
Connection health: GET /api/health returns {ok: true, v: 4.0.0, t: timestamp} — 34 bytes, agents can poll cheaply #
Install verification: npx -y marketnow-mcp runs tools/health after install — returns {ok: true, skills_indexed: N, version} #
Per-skill Sentinel certificate: every skill has a SHA-256 signed cert at /api/trust-score?skillId=X with recommendation: safe_to_install | install_with_caution | do_not_install
What is missing (roadmap): #
Telemetry opt-in — agents report connection success/failure to a public health dashboard #
Per-skill install success rate — like npm downloads but for actual worked/failed #
Failure taxonomy — structured enum so agents self-recover
Full agent hub: https://marketnow.site/for-agents MarketNow — the trust layer for agent commerce.
source & further reading
dev.to — original article
Re: --network none is your best friend — here's our full sandbox config + gVisor update
When AI Reviewers Disagree: Building a Multi-Agent DevSecOps Tribunal with Qwen-Max
How I Built a WhatsApp AI Assistant for a SA Salon - And What It Taught Me About Local AI