The Agent Skills Directory

[PROMPT_INJECTION]: The skill describes an iterative retrieval pattern that ingests external file content into an agent's context, creating a vulnerability surface for indirect prompt injection where malicious instructions hidden in the codebase could influence the agent's behavior.
Ingestion points: The retrieveFiles function and associated workflow in SKILL.md process file content from codebase paths like src/**/*.ts and routes/** based on user tasks.
Boundary markers: The pattern lacks instructions for the agent to use XML-like tags, delimiters, or explicit "ignore embedded instructions" warnings when processing the retrieved content.
Capability inventory: While the pattern is for retrieval, the skill is designed to be used by agents that perform downstream actions such as code modification, bug fixing, and feature implementation.
Sanitization: The logic does not include sanitization, validation, or filtering of the retrieved content before the agent evaluates its relevance and identifies context gaps.
[NO_CODE]: The skill consists entirely of markdown documentation and logic patterns without providing any executable scripts, binaries, or configuration files that run code directly.

iterative-retrieval