| name | escalation |
| description | Cap or restore the escalation tier for this conversation. Controls how far the orchestrator can escalate from frontdoor to specialist models. |
| version | 1.0.0 |
| metadata | {"hermes":{"tags":["routing","orchestrator","escalation"]}} |
/escalation — Cap Escalation Tier
Control how far the orchestrator can escalate requests through the model tier chain.
Usage
/escalation off — Frontdoor only, no escalation (tier A)/escalation B1 — Allow escalation up to coder tier/escalation B2 — Allow escalation up to architect tier/escalation full — Remove cap, full escalation chain (default behavior)
API Mapping
| Command | API Parameter | Value |
|---|
/escalation off | x_max_escalation | "A" |
/escalation B1 | x_max_escalation | "B1" |
/escalation B2 | x_max_escalation | "B2" |
/escalation full | (remove x_max_escalation) | — |
Escalation Tiers
A - Frontdoor only
B1 - + Coder escalation
B2 - + Architect escalation
C - + Worker specialist (full chain, default)
Notes
- Override parameters must be passed as strings
- Capping escalation reduces latency but may reduce quality on hard tasks
- Current model assignments are registry-owned; verify live roles via
GET /v1/models
- Use
/escalation off for quick factual queries where speed matters most
