SKILL PACK · FINOPS
Watch the bill. Cap the burn.
Cloud cost + LLM token spend in one console. Dynamic model routing keeps every agent call on the cheapest provider that meets the policy. Spike detection halts runaway agents before the invoice arrives.
WHAT'S INSIDE
Three sources of truth, one burn rate.
AWS, model providers, and your routing layer plug in as first-class MCP servers. Forge correlates every dollar back to the agent / pack / user that spent it.
APPROVAL TEMPLATES
Spend changes pass through you.
Four templates for the budget-shaping actions agents actually request. Each one writes a Guardian audit row with the dollar delta.
ANOMALY DETECTORS
Two ledgers watching burn velocity.
Streaming anomaly detection on cloud + token spend with 14-day rolling baselines. A 3σ jump triggers a kill-switch offer before the daily total reaches yesterday's.
- AWS daily $ vs week-ago
- GCP / Vercel cross-check
- Per-service spike detector
- Anthropic / OpenAI $ / hour
- Per-agent cost attribution
- Per-skill-pack chargeback
BUDGET CONTEXT
“Month-to-date spend trending 38% over plan. Auto-route Sonnet workloads to Haiku for 72h?”
FinOps Pack reads Kairon's billing-anomaly index alongside your routing config. Most agents are happy on a cheaper model; the pack proposes the swap and you swipe yes or no.
GOVERNANCE PROOF
Cost watched. Risk scored. Per pack.
Every paid tier ships the live governance console — the same one running over this page right now.
GOVERNANCE SCORE
- Cost VisibilityPer-provider spend + live burn rate aggregated into the Cost Radar.
- Agent AuditEvery agent action stamped with risk level and a 5-year retention log.
- GuardrailsRisk Console + emergency kill-switch wired to a kairon-guardian binary.
- Dynamic RoutingWorkloads are routed across providers by cost / quality / latency / context.
- Identity ProvenanceMCP token issued and at least one client (Claude Desktop / AEGIS / Cursor) connected.
COST RADAR · LIVE
Aggregate cloud-LLM spend across the monitored workload.
UNLOCK