smart-mode

Automating complexity assessment for autonomous development tasks involves scoring effort on a 1-5 scale to select an execution strategy. This skill evaluates task complexity to choose between DIRECT and ORCHESTRATED workflows, applying automatic scoring to determine the appropriate development mode.

Checkpoint-based failure recovery for multi-phase code execution uses a pivot, research, and checkpoint workflow to progress when blocked. It coordinates failure recovery strategies by establishing resume-from-checkpoint capabilities, allowing you to save state and continue execution without restarting.

Evidence gates in autonomous software workflows are mandatory verification steps requiring traceable evidence for every claim. They enforce evidence-based verification during task execution, ensuring that autonomous development modes validate progress before proceeding through checkpoints.

Use DIRECT execution modes for simple tasks and ORCHESTRATED modes for multi-phase work requiring checkpoints. The selection depends on automatic complexity scoring, where higher scores trigger ORCHESTRATED workflows to decompose work into phases and manage failure recovery.

Yes, you can integrate task decomposition and verification tools into autonomous execution. This skill reuses other skills like project-detector, verification-runner, and task-decomposer to provide end-to-end autonomy, coordinating complexity assessment with evidence gates.

Autonomous task execution fails to resume from the last checkpoint when evidence gates are not satisfied or the failure recovery workflow is incomplete. The system requires mandatory verification with traceable evidence before allowing checkpoint-based recovery to proceed.