Documentation / QueryTek Tapestry glossary / Context Channel
A context channel is the path along which the facts that make a request governable — who is acting, which organization they act for, what they are entitled to — travel with the request itself, rather than being re-established by each system that receives it.
Most integration failures are not transport failures. The message arrives; the context does not. The receiving system knows what was requested but not on whose authority, so it either re-derives that from its own copy of the data or assumes it.
A context channel treats that context as part of the payload's contract instead of an implementation detail. Three properties matter.
It is explicit. The context a receiver may rely on is declared rather than inferred from an endpoint, an API key, or a naming convention. Receivers can validate what they were given.
It is bounded. A channel carries what a downstream system legitimately needs to make its decision and no more. Passing broader context because it is convenient turns an integration into a data-sharing arrangement nobody agreed to.
It survives the hop. Context propagates through intermediate systems rather than being reconstructed at each one. Reconstruction is where two systems begin to disagree.
The distinction worth holding: a context channel is not the same as a correlation identifier. A correlation identifier lets you find related records after the fact. A context channel lets the receiving system decide correctly at the time, because it knows the organizational scope and entitlement the request carries.
When context is dropped, the cost surfaces later as reconciliation work — periodic jobs comparing systems that should never have diverged.
Whether eligibility and organizational scope travel with the work determines whether a cross-system process can be audited at all. If each system holds a partial view, an audit reconstructs intent from fragments and reconciliation becomes a standing operational task.
It is also a privacy control. Deciding deliberately what context crosses a boundary is what keeps an integration proportionate instead of a broad copy of workforce data moving because it was easier to send everything.
QueryTek Tapestry propagates eligibility, organizational scope, and progress with the request so downstream systems decide from current context rather than a local copy. Tenant boundaries remain enforced by the platform. Channel payload schemas, field-level mappings, and per-partner context contracts are established during onboarding and are not published.
