Multi-Agent Web Task Handoff Governance Framework

Role: Lead AI Systems Integrator specializing in multi-agent orchestration protocols and distributed web execution governance.

Context

• Multi-agent role topology: {{agent_role_topology}}
• Inter-agent payload schema: {{inter_agent_payload_schema}}
• Credential and security boundaries: {{credential_boundary_rules}}
• Human-in-the-loop escalation thresholds: {{human_in_the_loop_thresholds}}
• Shared memory and state store: {{shared_memory_protocol}}
• Chain termination criteria: {{chain_termination_criteria}}

Task

Design a multi-agent web task handoff governance framework that coordinates execution, standardizes cross-agent messaging, enforces security boundaries, and regulates handoff integrity across {{agent_role_topology}}.

Method

1. Deconstruct {{agent_role_topology}} to assign strict domain boundaries, distinct toolsets, and contextual scopes to each agent role.
2. Standardize an immutable handoff message structure using {{inter_agent_payload_schema}} to pass artifacts, URL states, and task statuses between agents.
3. Establish memory-scoping policies using {{shared_memory_protocol}} to maintain a clean distinction between global workflow history and agent-local execution logs.
4. Define security isolation rules enforcing {{credential_boundary_rules}} so privileged tokens are never leaked into downstream planner or extractor contexts.
5. Design pre-handoff verification gates where the dispatching agent must validate artifact completeness prior to relinquishing control.
6. Formulate deterministic escalation triggers based on {{human_in_the_loop_thresholds}} for financial, compliance, or destructive action requests.
7. Map terminal conditions and deadlock resolution strategies against {{chain_termination_criteria}} to ensure clean chain cessation.

Constraints

• Agents MUST NOT execute actions outside their explicitly designated role scope in {{agent_role_topology}}.
• Sensitive credentials MUST remain encapsulated in designated boundary nodes and never enter the raw message bus.
• Handoff payloads failing schema validation in {{inter_agent_payload_schema}} must be rejected immediately without execution.
• Workflows must halt and trigger human review upon crossing any condition defined in {{human_in_the_loop_thresholds}}.

Output format

Provide the complete governance architecture structured as follows:

1. Agent Role & Capability Blueprint (detailed breakdown of scope, permissions, and tool access per role in {{agent_role_topology}})
2. Inter-Agent Communication Protocol (JSON schema specification for handoffs, task tokens, and status transfers)
3. Security & Context Isolation Boundary (concrete isolation rules and credential protection mechanisms)
4. Handoff Verification Gates & Deadlock Engine (step-by-step handoff verification rules, escalation paths, and termination handlers)

Self-review

• Confirm that no agent role has unauthorized access to tools beyond its direct responsibility.
• Ensure the inter-agent payload schema fully supports all state transfers required by {{shared_memory_protocol}}.
• Validate that human escalation paths in {{human_in_the_loop_thresholds}} are unambiguous and non-bypassable.