The Agent Skills Directory

[COMMAND_EXECUTION]: The skill instructs the agent to execute a local bash script (resolve-rules.sh) located in the vendor's plugin root directory. This script is used to load and resolve custom rules for the brainstorming session from project or user-level configuration files.
[PROMPT_INJECTION]: The skill contains an indirect prompt injection surface as it ingests content from external files (.claude/brainstorm-rules.md and $CLAUDE_PLUGIN_DATA/brainstorm-rules.md) and treats it as supplemental instructions for the agent.
Ingestion points: Reads content from local markdown files via a bash script located in the vendor directory.
Boundary markers: No specific delimiters or safety instructions are defined to separate the external rule content from the skill's core system prompts.
Capability inventory: The skill possesses access to significant tools including Bash, Write, Edit, Agent, and Skill, which could be influenced by malicious instructions in the rules files.
Sanitization: Loaded content is interpreted directly as natural language instructions without validation or sanitization.

brainstorm