Built-in Roles¶

Source of truth: src/assync/roles/builtin.yaml, loaded by roles/registry.py. Generated directly from the shipped registry — 27 executable roles plus one non-executable entry. See Role Pool for the model and validation rules.

Governance¶

ID	Name	Risk	Executable
`governance_authority`	Human Governance Authority	critical	No
`project_manager`	Project Manager	low	Yes
`solution_architect`	Solution Architect	medium	Yes
`technical_reviewer`	Technical Reviewer	medium	Yes
`agent_auditor`	Agent Auditor	medium	Yes

Development¶

ID	Name	Risk	Executable
`lead_programmer`	Lead Programmer	medium	Yes
`backend_engineer`	Backend Engineer	medium	Yes
`frontend_engineer`	Frontend Engineer	medium	Yes
`ai_engineer`	AI Engineer	medium	Yes
`devops_engineer`	DevOps Engineer	high	Yes

Quality¶

ID	Name	Risk	Executable
`security_engineer`	Security Engineer	high	Yes
`qa_engineer`	QA Engineer	medium	Yes
`performance_engineer`	Performance Engineer	medium	Yes
`reliability_engineer`	Reliability Engineer	medium	Yes

Specialist¶

ID	Name	Risk	Executable
`documentation_engineer`	Documentation Engineer	low	Yes
`ux_designer`	UX Designer	low	Yes
`data_engineer`	Data Engineer	medium	Yes
`release_manager`	Release Manager	high	Yes
`compliance_officer`	Compliance Officer	medium	Yes

AI System¶

ID	Name	Risk	Executable
`agent_systems_engineer`	Agent Systems Engineer	medium	Yes
`memory_engineer`	Memory Engineer	medium	Yes
`prompt_engineer`	Prompt Engineer	low	Yes
`conversation_engineer`	Conversation Engineer	low	Yes
`knowledge_engineer`	Knowledge Engineer	low	Yes
`mcp_engineer`	MCP Engineer	medium	Yes
`automation_engineer`	Automation Engineer	medium	Yes
`ai_safety_engineer`	AI Safety Engineer	high	Yes

Human Governance Authority¶

governance_authority is the one non-executable entry. It represents the Human Mission Commander for separation-of-duty checks and authority-model references — roles/validation.py rejects any attempt to make it executable.

Inspect a role's full definition¶

assync role show <role_id>

Prints the role's description, responsibilities, allowed/forbidden capabilities, required evidence, conflicts_with, default adapter, and risk level.