The Agent Skills Directory

[COMMAND_EXECUTION]: The skill uses the Bash tool to execute Python scripts, passing the user-supplied directory path (args) as a command-line argument to analyze the codebase.
[DYNAMIC_EXECUTION]: The skill dynamically generates Python code via shell heredocs to interact with the trailmark API, facilitating complex structural analysis and graph generation from a fixed template.
[EXTERNAL_DOWNLOADS]: The skill uses uv run --with trailmark to execute its logic. This command may download and install the trailmark package from a public registry if it is not locally available. This package is a resource provided by the skill's author (trailofbits).
[INDIRECT_PROMPT_INJECTION]: The skill analyzes untrusted codebases, creating a potential surface for indirect prompt injection.
Ingestion points: Files within the target directory specified by the args parameter in SKILL.md are processed by the analysis engine.
Boundary markers: No explicit markers are defined; the skill processes all files detected as supported languages within the target scope.
Capability inventory: Access to Bash, Read, Grep, and Glob tools allows for command execution and filesystem access during the analysis process.
Sanitization: The skill relies on the trailmark library's internal parsing and does not perform additional sanitization of the code content before analysis.

trailmark-structural