ghchinoy/scion-platform
Scion platform administration, Hub bootstrap and GCE reconciliation, authorization and broker quota diagnostics, zero-downtime agent token re-minting, fleet model assessment, live model patching, and server lifecycle operations.
Audit fleet model and Vertex AI region configurations across Scion Hub projects, detect static env overrides (ANTHROPIC_MODEL, CLOUD_ML_REGION) and unpinned fallback drift, and live-switch running agent models. Requires jq and curl.
Diagnose Scion Hub AK1 authorization denials (403 Forbidden, CanDelegate, missing scopes) and broker quota ceilings (429 Too Many Requests, max_agents_per_broker), inspect role definitions and delegation ceilings, and execute zero-downtime agent token re-minting via Hub admin APIs. Requires the scion CLI, jq, and curl.
Bootstrap, upgrade, and reconcile self-hosted Scion Hub deployments on GCE, including clean database resets, selective container image builds (claude, antigravity, opencode), headless super-admin JWT derivation, Vertex AI global region configuration, broker quota ceilings, and global agent-team template imports.