The Agent Skills Directory

[COMMAND_EXECUTION]: The skill interpolates the {args} parameter directly into shell command strings within SKILL.md across multiple steps. This practice allows for potential command injection if the input provided to the {args} variable contains shell metacharacters such as semicolons, pipes, backticks, or command substitution syntax.
[DYNAMIC_EXECUTION]: The skill utilizes dynamic execution in SKILL.md by using uv run --with trailmark to resolve, install, and execute dependencies at runtime. Furthermore, it executes logic by piping a multi-line Python script into the interpreter via a shell heredoc (<<'PY') in Step 2.
[INDIRECT_PROMPT_INJECTION]: The skill ingests and processes untrusted data from external codebases, creating a vulnerability surface for indirect prompt injection.
Ingestion points: The trailmark tool in SKILL.md processes directory structures and file contents from the target path provided in {args}.
Boundary markers: None. The instructions do not define delimiters or provide warnings to the agent regarding instructions potentially embedded in the analyzed code.
Capability inventory: The skill uses Bash for command execution and has file system access via Read, Grep, and Glob tools.
Sanitization: There is no sanitization or validation performed on the input arguments or the codebase data before processing.

trailmark-summary