executing-plans

name	executing-plans
description	Use when you have a written implementation plan to execute inline in the current session with review checkpoints. Prefer subagent-driven-development when sub-agent dispatch is available.
version	1.0.0
author	Surogate Agent (adapted from obra/superpowers)
license	MIT
tags	["process","implementation","execution"]

Executing Plans

Overview

Load plan, review critically, execute all tasks, report when complete.

Announce at start: "I'm using the executing-plans skill to implement this plan."

Note: This skill drives task-by-task execution inline in the current session. If your platform supports sub-agent dispatch (delegate_task, spawn_worker), the subagent-driven-development skill produces higher-quality results with two-stage review and isolated contexts. Use this skill when sub-agent dispatch is not available or your human partner explicitly chose inline execution.

The Process

Step 1: Load and Review Plan

1. Read the plan file with read_file.
2. Review critically — identify any questions or concerns about the plan.
3. If concerns: raise them with your human partner before starting.
4. If no concerns: create a todo per task and proceed.

Step 2: Execute Tasks

For each task:

1. Mark the todo as in_progress.
2. Follow each step exactly (the plan has bite-sized steps).
3. Run verifications as specified — don't skip them.
4. Mark the todo as completed only when the step's verification passes.

Step 3: Complete Development

After all tasks complete and verified:

• Announce: "All plan tasks complete. Tests pass. Ready to wrap up."
• Run any final-stage verification the plan demands (full test suite, lint, type checks).
• Ask your human partner how they want to integrate the work (merge, PR, hand off).
• Do not push, merge, or open a PR without explicit confirmation.

When to Stop and Ask for Help

STOP executing immediately when:

• You hit a blocker (missing dependency, test fails, instruction unclear).
• The plan has critical gaps preventing the next step.
• You don't understand an instruction.
• A verification fails repeatedly after the same fix attempt.

Ask for clarification rather than guessing.

When to Revisit Earlier Steps

Return to Review (Step 1) when:

• Your human partner updates the plan based on your feedback.
• The fundamental approach needs rethinking.

Don't force through blockers — stop and ask.

Remember

• Review plan critically first.
• Follow plan steps exactly.
• Don't skip verifications.
• Reference skills when the plan says to (use skill_view).
• Stop when blocked — don't guess.
• Never start implementation on the main/master branch without explicit consent.

Integration

Related skills:

• writing-plans — Creates the plan this skill executes.
• subagent-driven-development — Higher-quality alternative when sub-agent dispatch is available.
• systematic-debugging — Use when a verification fails and the cause isn't obvious.