The Agent Skills Directory

[PROMPT_INJECTION]: The skill exhibits an Indirect Prompt Injection surface. It ingests untrusted user requirements and design specifications to generate a comprehensive markdown implementation plan. This generated plan is specifically intended to be executed by other agents (referencing superpowers:executing-plans and superpowers:subagent-driven-development).
Ingestion points: User-provided design and feature requirements are processed to populate the implementation tasks in SKILL.md.
Boundary markers: There are no explicit delimiters or instructions provided to the agent to ignore or escape potentially malicious commands embedded within the user's design input.
Capability inventory: While this skill itself only writes files, the output is explicitly designed to drive tools and subagents with command execution and file modification capabilities (via the handoff instructions).
Sanitization: No sanitization or validation of the input data is performed before it is interpolated into the code examples and command blocks of the generated plan.

writing-plans