The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill instructs the agent to fetch a curated list of tools and resources from an external GitHub repository: https://raw.githubusercontent.com/gmh5225/awesome-llvm-security/refs/heads/main/README.md. While the source is a standard reference for LLVM security, it originates from a third-party account not explicitly identified as a trusted vendor.
[INDIRECT_PROMPT_INJECTION]: The practice of fetching and processing remote markdown content creates a vulnerability surface where instructions embedded in the external README could influence the agent's logic or behavior.
Ingestion points: The file references/integration-and-resources.md directs the agent to retrieve data from a remote URL.
Boundary markers: There are no instructions provided to the agent to treat the fetched content as untrusted or to ignore embedded instructions.
Capability inventory: The skill provides instructions for using system-level tools like clang, perf, and various LLVM utilities, which could be misused if the agent is influenced by malicious content in the fetched file.
Sanitization: No validation or sanitization mechanisms are defined for the external content before it is processed by the agent.

dynamic-instrumentation