Method
Agentic Workflow Rail
Builds a swimlane control map for user, agent, policy, memory, retrieval, tools, human approval, audit and rollback.
Field scoring contract
| Field | Type | Role | Direction | Scored |
|---|---|---|---|---|
| Agent mission | text | context | context | no |
| Memory use | select | risk_driver | higher_is_worse | yes |
| Retrieval scope | select | risk_driver | higher_is_worse | yes |
| Tool access | select | risk_driver | higher_is_worse | yes |
| Approval model | select | risk_driver | higher_is_worse | yes |
| Rollback path | select | risk_driver | higher_is_worse | yes |
Limits
- Preliminary output
- Human review required
- Not certification
- Risk/readiness separated where implemented
- Context text is not averaged into numeric scores.
- Outputs require source/evidence review before decisions.