Subagent orchestration and verification discipline. Use whenever a task has independent subtasks that could run in parallel, whenever work is substantial enough to need checking before it ships, and whenever tempted to review your own work by re-reading it. Trigger on multi-file changes, audits, migrations, reviews, and any long-running build.
Data-skepticism discipline for quantitative work. Use whenever reading data and producing a conclusion: analytics questions, metrics reports, spreadsheet analysis, SQL queries, dashboards, A/B results, forecasts, or any claim of the form "the data shows." Trigger before presenting any number or trend to a user.
Autonomy and checkpoint discipline for agentic work. Use on every task where the model works with tools over multiple steps. Trigger especially when about to ask the user a question, end a turn, propose a plan, or write phrases like "I'll now," "next I will," "want me to," "shall I," or "let me know."
Cross-run memory discipline for agents with a persistent workspace. Use at the end of any session where something non-obvious was learned (a correction from the user, a confirmed approach, a trap in the environment), and at the start of any task similar to past work. Trigger on user corrections, surprising failures, and hard-won configuration or debugging discoveries.
Scope discipline for code changes. Use whenever writing or editing code: implementing a feature, fixing a bug, or modifying existing files. Trigger especially when tempted to refactor, clean up, add error handling, add configuration options, or restructure code beyond what the task asked for.
Prose discipline for any written deliverable. Use whenever writing for a human reader: answers, explanations, reports, summaries, documentation, PR descriptions, emails, or analyses. Trigger especially when the draft is getting long, when reaching for headers and bullet points, or when compressing text with fragments and abbreviations.
Final-summary discipline for agentic sessions. Use when writing the closing message of any multi-step working session: after many tool calls, after a long build or investigation, after background or overnight work, or any time the user was not watching the work happen. Trigger on every end-of-task summary.
Debugging discipline. Use whenever diagnosing a bug, error, test failure, crash, regression, flaky behavior, or unexpected output, and especially before applying any fix. Also trigger when a previous fix attempt did not work and another attempt is about to be made.