nik1t7n

Use when planning, reviewing, or improving research work: choosing research problems, designing ML/AI experiments, reading papers, building experiment loops, analyzing outputs, writing research notes, or turning vague curiosity into a disciplined research plan. Emphasizes avoiding absorbed problems, upgrading information inputs beyond shared reading lists, and reasoning backward from outcomes the researcher genuinely wants to exist.

Use when the user wants deterministic multi-agent workflows, feedback loops between builder/reviewer agents, launch-control style implementation gates, repeatable Pi subagent processes, or custom workflow construction with max iterations, gates, stop conditions, and artifacts.

Use when implementing a medium or large coding project from a written spec, plan, or product goal where drift, skipped validation, PR correctness, or multi-block delivery risk matter.

Design and operate efficient multi-agent coding DAGs for medium or large software projects. Use when Codex needs to decompose a specification into parallel executor leaves, define dependencies and integration gates, choose GPT-5.6 models, isolate worktrees, control executor-reviewer loops, prevent graph drift, use risk-based validation, maintain deterministic run state, or recover a long-running DAG that is slow, expensive, cyclic, or blocked.

Use when running noisy terminal commands (tests, builds, wide searches, diffs, large logs) that might flood the agent context window with useless lines.