observer

Harness for visible, reviewable, resumable agent-run data movement. Use when writing, adapting, or running a pipeline that transforms, enriches, imports, exports, syncs, backfills, sends, or writes records; when the job spends API credits or mutates a CRM, database, spreadsheet, file, webhook, or API; when adding dry-run approval, source locks, durable resume, JSONL ledgers, dashboard rows, run controls, or dashboard chat; or when maintaining Observer Kit itself.

Orchestration playbook for multi-stage agent-run data transformations. Use when chaining scripts or API steps, making fields depend on earlier results, building a DAG or Clay-style workflow, conditionally enriching rows, joining or expanding entities, rerunning affected steps, caching node outputs, creating a user cookbook of reusable nodes, subflows, or adapters, or coordinating several Observer Kit operations as one reviewed flow.