The Agent Skills Directory

[PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection as it clones and reads files from external repositories. Maliciously crafted documentation or comments within a target codebase could attempt to subvert the agent's logic.
Ingestion points: SKILL.md instructions clone target repositories to /tmp/target-repo and read their content.
Boundary markers: The instructions lack explicit delimiters or warnings to ignore embedded instructions found within the analyzed codebase.
Capability inventory: SKILL.md uses shell commands including ls, find, and wc to inspect the file system.
Sanitization: No content validation or sanitization of the cloned files is specified.
[EXTERNAL_DOWNLOADS]: The skill uses git clone to fetch repositories from GitHub. This is a well-known service and the operation is consistent with the skill's stated purpose.
[COMMAND_EXECUTION]: The skill instructs the agent to execute common shell commands (ls, find, wc, git) to analyze the target repository. These commands are used for inspection purposes and do not involve executing code from the downloaded repositories.

software-architecture-analysis