talk-pipeline

Automating talk preparation requires an orchestrator to sequence stages from extract to revision, enforcing stage order and validating checkpoints. This process coordinates context collection, stage routing, and output generation across multiple sub-skills.

Stage routing in a talk preparation pipeline is the mechanism that directs processing flow across sequential stages like extract, research, concepts, position, script, and revision. It supports both REX and Concept modes while validating human-in-the-loop checkpoints.

Yes, you can run individual pipeline stages per-stage as needed rather than executing the full end-to-end talk preparation sequence. The orchestrator supports both full sequential processing and targeted per-stage execution with built-in checkpoints.

The orchestrator enforces standardized output file naming per the pipeline guide to simplify sharing and publishing of talk preparation artifacts. This ensures consistent artifact generation across extract, research, concepts, position, script, and revision stages.

REX and Concept modes are supported natively by the orchestrator, providing clear routing and checkpoints without manual configuration. The pipeline automatically coordinates mode-specific context collection and stage routing across all sub-skills.

Automating talk preparation orchestration with human-in-the-loop checkpoints requires manual validation at each stage before proceeding, meaning full automation is intentionally interrupted. This prevents fully unattended runs but reduces errors in stage sequencing and output generation.