The Domain-Specific Agent Harness
Answer summary
A skill tells an agent how to do one task. A domain-specific harness composes playbooks into a persona, runs them against a goal, and holds the irreversible step for a person.
What is a domain-specific agent harness? A domain-specific agent harness is the shared contract around agent skills. Playbooks declare inputs, steps, outputs, connectors, and approval rules. A persona groups those playbooks into teams, and a goal invokes the playbooks a profession needs. A person approves the irreversible step.
How is a playbook different from an agent skill? A skill gives an agent reusable instructions, and sometimes scripts or reference files. A playbook is the contract for how that work runs with other work: declared inputs and outputs, connector operations, and approval rules. The workspace binds the capabilities when the playbook runs, so another team can reuse the same definition.
Where is the Domain-Specific Agent Harness Spec? The open specification is published at https://gabriel-operator.github.io/domain-specific-agent-harness-spec/. Draft 0.1 covers the overview, the specification, apps and connectors, worked examples, and how the draft relates to Agent Skills, MCP, and A2A.
Published as a Technology story in November 14, 2026.
This article is part of the official Gabriel Operator publication about AI personas, governed operators, browser automation, and agent infrastructure.