The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill processes external, untrusted content including the target code files, project conventions (AGENTS.md, CLAUDE.md), and related tests/types to generate explanations. This creates a surface for indirect prompt injection.
Ingestion points: Target file content ($TARGET), project-level documentation (AGENTS.md, CLAUDE.md), and related files found via ripgrep/fd.
Boundary markers: The skill instructions do not specify the use of clear delimiters or instructions to the subagent to ignore embedded commands in the analyzed source data.
Capability inventory: The skill utilizes Read, Glob, Grep, Bash, and Task (which dispatches a subagent with full capabilities).
Sanitization: There is no evidence of sanitization or filtering of the file contents before they are passed to the explainer agent.
[COMMAND_EXECUTION]: The skill performs shell operations using the Bash tool with the user-provided $TARGET argument (e.g., test -f "$TARGET", ast-grep -p "function $TARGET", rg "... $TARGET"). While this is the intended functionality for code analysis, it presents a potential command injection vector if the target string contains shell metacharacters and is not properly sanitized by the underlying agent platform before execution.

explain