增强版写作计划
PassAudited by ClawScan on May 14, 2026.
Overview
This is an instruction-only planning template with no code or credentials; the main caveat is that its saved plans can steer later agent execution.
This appears safe to use as a planning aid. Before reusing or committing generated plan files, review the header that tells Claude to use executing-plans, and choose subagent execution only when you are comfortable with the scope and review checkpoints.
Findings (2)
Artifact-based informational review of SKILL.md, metadata, install specs, static scan signals, and capability signals. ClawScan does not execute the skill or run runtime probes.
A later agent reading the saved plan may treat this as an instruction to invoke an execution skill rather than merely using the plan as documentation.
The generated plan is required to contain an agent-directed instruction that can force a future Claude session toward a specific execution skill.
> **For Claude:** REQUIRED SUB-SKILL: Use superpowers:executing-plans to implement this plan task-by-task.
Keep this line only if you want future sessions to use that execution workflow; otherwise remove it or reword it as an optional suggestion requiring user confirmation.
If you choose this mode, task details and relevant project context may be passed to subagents for implementation.
The handoff offers a subagent-based execution mode. It asks the user to choose and includes review between tasks, but it does not define the exact context or permissions each subagent receives.
Subagent-Driven (this session) — I dispatch fresh subagent per task, review between tasks
Use subagent execution only after confirming the task scope and checkpoints, especially for sensitive repositories or high-impact changes.