
One AI Step Failed. What Else Does It Block?
Use NetworkX to map mandatory workflow dependencies, reject cycles and find the downstream tasks blocked by a failure. Includes a seven-task example and twelve runtime tests.
BY JASON EAD
Original guides and analysis about coding tools, models, GitHub projects, agents, and creative AI workflows.

Use NetworkX to map mandatory workflow dependencies, reject cycles and find the downstream tasks blocked by a failure. Includes a seven-task example and twelve runtime tests.

Catch duplicate lookup keys and unmatched accounts before a data join multiplies your AI workflow records. Reproduce the pandas example with 12 runtime tests.

Build a tested Jinja rendering boundary that rejects missing report fields, escapes HTML text and keeps human approval separate from delivery.

Build a local Datasette console to inspect AI workflow records, filter review queues and understand the limits of read-only snapshots and export controls.

Build a local DOCX-to-Markdown review bundle with source snapshots and hashes. A tested example shows why successful conversion can still lose business context.

Build a local review queue for possible duplicate CRM companies. Includes tested RapidFuzz code, a fictional dataset, threshold results and limits for AI workflows.

Build a tested local DuckDB workflow that calculates weekly order metrics, produces a reproducible JSON brief and gives an AI assistant a focused business-review task.