neutral data layer

Embed the reliability feed // gateways & routers

Building an MCP gateway, tool router, or agent platform? Power your routing with a neutral, cross-ecosystem reliability signal — without building the data pipeline yourself. Your gateway sees only your traffic; Glimind sees across the whole ecosystem.

The neutral feed

# cacheable snapshot: open breakages + recent recoveries + per-tool scores
curl https://glimind.com/v1/reliability/feed        # poll up to 1x/min

# on-demand batch for the exact tools a request touches
curl "https://glimind.com/v1/health?tools=github/search,slack/post,stripe/charge"

How to use it in routing

Pre-flight: before dispatching, check the target's verdict; if it's down, route to a healthy alternative immediately instead of failing and retrying.
Failover ranking: rank candidates by Glimind score so your fallback picks the substitute that's actually healthy across the ecosystem — not just the last one you tried.
Corroboration gate: corroborated: true means ≥2 independent reporters agree — use it to avoid flapping on a single noisy signal.

Why embed instead of build

Deriving reliability from your own logs is single-tenant: blind to outages until they hit your traffic, and biased to the tools your customers already use. Glimind is the independent rating layer — cross-ecosystem, neutral (we sell no tools and run no integrations), and citable. You stay the execution path; we're the data beneath it.

Two-way (optional, recommended)

Forward privacy-clean outcomes of the calls you broker (via the SDK or an OTLP export) and your tools' scores get sharper while you earn higher limits. Fully payload-free — see privacy.

Partner / volume access: get a key · full reference.