The Agent Skills Directory

[PROMPT_INJECTION]: The skill implements a learning mechanism that stores potentially unvalidated data into a reference file, creating a surface for indirect prompt injection.
Ingestion points: The agent reads references/learnings.md at the start of each session.
Boundary markers: No delimiters or instructions are used to separate the learned content from the agent's primary instructions.
Capability inventory: The skill can route to other tools and provides instructions for the agent to append new information to the references/learnings.md file.
Sanitization: No sanitization or validation of the content written to the learnings file is specified.
[COMMAND_EXECUTION]: The skill documentation includes an example command npx skills add sales-skills/sales --skill sales-do -a claude-code -y for installing related tools from the same vendor. This is presented as an example for users to extend their environment.

sales-oliv