Anthropic provides
The harness is the agent loop itself — Claude reasoning, choosing tools, acting, and iterating until the task is done. In Claude Managed Agents, Anthropic hosts and operates this loop on their infrastructure.
Tasks, tools, and guardrails are defined through the API or the Claude Console; the platform manages state, permissions, and execution so teams ship the agent, not the plumbing.
Skoor adds
Skoor adds the certified profile that constrains the loop: which connectors, which scopes, which refusals — versioned, signed, and applied on your approval. A configuration you did not agree to is a supply-chain risk wearing a helpful face, so profiles are never pushed silently.
An agent that runs inside the ring is a skoored agent: profile attached, scopes declared, skoorboard row created at birth, its Agent Skoor starting empty and earned from conduct. Manufactured governed, not retrofitted.
Stated plainly
The profile is a specification Skoor certifies — the loop itself is Anthropic's, and Skoor never operates or proxies it.
The assessment establishes what your agents can reach at every edge — and produces your baseline Company Skoor.