Request observability

Purpose

Syndicate Code exposes request-centric observability so operators can pivot from a checkpoint or audit question to the exact gateway execution history for that request.

This is not a second enforcement path. Enforcement still happens at the control plane and execution boundary. Observability provides correlated evidence after and during execution so teams can investigate what happened.

Correlation model

The observability model is built around a shared request_id.

• gateway telemetry emits accepted, completed, failed, and stream-finalized events keyed by request_id
• checkpoint projections preserve the same request_id
• operator query surfaces resolve by request_id rather than forcing direct event-store or database access

This lets an operator move from checkpoint review to execution evidence without reconstructing joins manually.

Operator surfaces

Syndicate Code now exposes request observability through these interfaces:

These surfaces are intentionally read-focused. They help with incident review, support triage, and audit evidence collection.

GraphQL correlation

The GraphQL API remains the primary read surface for interactive clients. For observability, two request-level queries matter:

• checkpointPending, which now preserves requestID
• gatewayEventsByRequestID(requestID, limit), which returns canonical gateway event records for the specified request

This keeps TUI and automation clients on the same evidence model instead of inventing separate ad hoc APIs for request tracing.

CLI workflow

Typical operator flow:

syndicate telemetry events --request-id req_123 --output table
syndicate telemetry events --request-id req_123 --output json

The CLI is the fastest path for support and compliance workflows where an operator already has a request ID from a checkpoint, audit review, or customer report.

Relationship to Syndicate Gate

When Syndicate Code runs through Syndicate Gate, the shared request_id links:

• the Code-side checkpoint and control-plane state
• the Gate-side inference and routing events
• deployment-specific dashboard surfaces used by hosted and enterprise operators

That cross-product linkage is what turns raw logs into usable evidence.