| name | workflow/escalate-to-human |
| description | Escalates a session issue to a human reviewer. |
| version | 0.1.0 |
| status | experimental |
| family | workflow |
| role | Strategist |
| allowed_tools | ["list_dir","write_to_file"] |
| policies | ["E.16 RoC-Autonomy","A.15.1 U.Work"] |
| outputs | ["EscalationRecord"] |
Workflow: Escalate to Human
1. Context
This skill records a human escalation request for a session issue, capturing severity, target, and related artifacts.
2. Inputs
- context (required): Bounded context name (safe path segment).
- session_id (required): Session identifier (safe path segment).
- reason (required): Escalation reason or trigger summary.
- severity (optional): Escalation severity (
low, medium, high).
- target (optional): Escalation target (default: human).
- requested_by (optional): Requester label.
- artifacts (optional): Semicolon-delimited artifact paths.
- notes (optional): Freeform notes.
- agent_model (optional): Agent model identifier.
- role_assignment (optional): RoleAssignment for U.Work logging (default: Strategist).
- decisions (optional): Semicolon-delimited DRR ids/paths.
- timestamp_start (optional): ISO-8601 timestamp for deterministic output.
3. Outputs
runtime/contexts/<context>/escalations/<session_id>.<n>.escalation.md
4. Procedure
- Validate inputs and compute the next escalation index.
- Write the escalation record with status open.
- Emit U.Work via
telemetry/log-work when available.