Guides

Configure NEXUS, FORGE, SENTINEL, LORE, and VESSEL - registry.json, model routing, Coder modules, and worker scaling.

Per-agent provider selection, cost optimization, adversarial review across model families, and AI Gateway fallback.

Shrink FORGE context, reduce SENTINEL rounds, downgrade VESSEL/LORE models, scale instances deliberately.

Air-gapped deployments: configure LiteLLM proxy for local models when Coder AI Gateway is unavailable.

Tenant model, Coder RBAC, Redis keyspace isolation, and per-tenant NEXUS workspaces.

Commands to add, remove, isolate, and resume tenants.

Per-tenant NEXUS workspace setup, fault isolation, and service accounts.

Add skills via SKILL.md, register in registry.json v2.

Register MCP servers globally or per-agent in registry.json.

Custom model definitions, aliases, cost tiers in registry.json.

Production runbooks, TUI usage, log inspection, Redis monitoring, and scaling workers.

Common failures, diagnostics, Coder connectivity, token issues, and how to read STATUS.json.

Coder API reachability, token rotation, GitHub external auth issues.

Worker timeouts, SENTINEL rejection loops, merge conflicts, STATUS.json deep-dive.