Skip to content
Back to skills

Executing Plans

ASecurity

Execute an existing implementation plan through verification. Honor review checkpoints explicitly requested by the user or plan.

  • 10 stars
  • 0 votes
  • 0 copies
  • 0 views
  • Added September 23, 2026
ai-agentsapi

Works with

  • api

Security analysis

A100/100

Scanned September 23, 2026

npx -y skills add shekohex/dotai --skill executing-plans --agent claude-code

Installs into .claude/skills of the current project.

Are you the author of Executing Plans?

Add the live security badge to your README. It updates with every re-scan.

Security grade badge for Executing Plans
[![Security: A — Skills Directory](https://www.skillsdirectory.com/api/skills/shekohex-executing-plans/badge)](https://www.skillsdirectory.com/skills/shekohex-executing-plans)

More formats (shields.io, HTML) on the badges page. Keep it an A: scan every change in CI with Pro.

Download with Pro
SKILL.md
---
name: executing-plans
description: Execute an existing implementation plan through verification. Honor review checkpoints explicitly requested by the user or plan.
---

# Executing Plans

Carry an approved or user-supplied implementation plan through its acceptance criteria.

1. Read the plan and relevant repository guidance. Check dependencies, intended outcome, verification steps, and any explicit review checkpoints.
2. Resolve routine details from code and context. Ask only about gaps that change scope, correctness, or authorization; continue independent tasks meanwhile.
3. Execute in dependency order. Track meaningful tasks, adapt stale mechanics to current APIs, and preserve the intended behavior.
4. Run affected checks and required repository gates. Diagnose and fix in-scope failures, then rerun affected checks.
5. Report progress at useful milestones and continue until all authorized tasks are complete. Pause between batches only when the user or plan explicitly requires review there.

When a blocker cannot be resolved within scope, report the exact missing input or access and work already completed. A failed test is evidence to investigate, not an automatic handoff.

Done means plan requirements are implemented and verified, with remaining limitations identified. Commits, publishing, and deployment follow existing authorization; executing a plan does not independently authorize them.

Attribution

Is this your skill, or is something wrong with this listing? Request removal or report an issue. Author removals are honored within 72 hours.

Comments

Loading comments…

Newsletter

Get the weekly list of safe new skills, and the ones we pulled.