Observing the Pipeline
Every pipeline run produces a stream of structured events — stage transitions, agent decisions, security verdicts, and resource consumption metrics. The monitoring system captures these events in real time, making the pipeline's internal reasoning transparent and auditable.
Progress Tracking
The CLI displays a live progress indicator during pipeline execution. Each stage reports its status — queued, in-progress, completed, or failed — along with the elapsed time and a brief summary of the outcome. For multi-chain deployments, the progress tracker displays parallel stage execution across target chains.
Cost Metrics
The cost guardian tracks expenditure in real time across three dimensions: AI model inference costs, blockchain gas fees for deployment, and platform compute time. Each pipeline run produces a cost breakdown that maps spending to individual stages.
When a run approaches the configured budget ceiling, the pipeline pauses and requests authorisation before continuing. This prevents runaway costs during iterative development — particularly important when working with large models or deploying to multiple chains simultaneously.
Audit Trail
Every pipeline run generates a complete audit record. The record includes the input specification, the agent swarm's deliberation log (including dissenting opinions and the security agent's analysis), the optimisation decisions applied, and the final output hash. Audit records are immutable and addressable by a unique run identifier.
Audit trails are especially valuable for regulated industries. The complete record of agent reasoning — including which security checks were applied and why certain optimisation strategies were chosen or rejected — provides the documentation trail that compliance teams require.
The audit trail is accessible through both the CLI (ludopoly audit <run-id>) and the web IDE's run history panel. Records can be exported in JSON format for integration with external compliance and reporting systems.