Yellowhead Digital documentation

How the Yellowhead Digital platform works for the people — and the agents — that operate it.

Yellowhead Digital is a forensic marketing-infrastructure diagnostic platform. These docs cover the parts of the platform that a technical reader or an autonomous agent needs to integrate with: the agent door (a standards-based Model Context Protocol (MCP) server), the auth and capability model that governs it, and the platform surfaces those tools read and write.

Who this is for

• Agent and integration builders — teams connecting an LLM agent (Cursor, Claude Desktop, a custom workflow) to a Yellowhead workspace over MCP.
• Technical evaluators — anyone assessing how the platform authenticates agents, scopes their authority, audits their actions, and holds their writes.

Start here

• The agent door — what it is, its live status, and how access is granted.
• Authentication — ID-JAG, OAuth discovery, and DPoP.
• Capabilities & grants — how an agent's authority is scoped.
• Tools reference — the tools the door exposes today.
• Governance & audit — RLS, the per-call audit trace, approvals, anomaly alerts, and rate limits.
• Reports & exports — report data and the export formats available today.

Two doors, one source of truth

Every record in the platform lives in one place. A human reaches it through the dashboard; an agent reaches the same rows through the agent door. There is no sync layer and no second copy of the truth — the same row-level security that scopes a person scopes an agent. The agent-door docs describe the second door in detail.