agent-coding-workflows
Choose and run the standard AI coding workflows by chaining the other skills; use when the user asks for a workflow or uses /flow, /workflow, /andrej, or /AndrejHide, and when the task type is standard change, algo/perf, or urgent bugfix.
assumption-clarifier
Identify hidden assumptions, contradictions, and missing info before coding; use when requirements are vague, ambiguous, or risky, and when a task needs clarification questions or explicit success criteria.
conceptual-reviewer
Review code for conceptual errors, wrong assumptions, edge cases, and overcomplication; use after medium/large changes or when risk is high.
criteria-test-loop
Drive coding by success criteria and tests; use for bugfixes, features with tests, and algorithmic changes. Emphasize naive-correct first, then optimize.
minimal-diff-implementer
Implement the smallest safe change and avoid bloat; use for most coding tasks to prevent overengineering, unrelated edits, or destructive refactors.
plan-lite
Create a lightweight 3-5 step plan with risks and checkpoints; use when a task is non-trivial, has regression risk, or spans multiple files/systems.