The Agent Skills Directory

[COMMAND_EXECUTION]: The skill provides a series of shell commands in the 'Quick Commands for Fingerprinting' section. These commands are used for local project analysis and do not involve remote execution or system-wide modifications. Specifically, they use common tools like cat, ls, git, find, and wc to gather project metadata.
[DATA_EXPOSURE_&_EXFILTRATION]: While the skill reads project configuration files such as package.json and tsconfig.json, it does so locally to generate a project profile. There are no network operations detected that would exfiltrate this data to an external server.
[INDIRECT_PROMPT_INJECTION]: The skill processes project structure and configuration files which could theoretically contain malicious instructions. However, the skill treats this data as structured input for classification purposes rather than executing it as instructions.
Ingestion points: Files like package.json, tsconfig.json, and directory listings via ls (SKILL.md).
Boundary markers: None explicitly defined in the provided markdown.
Capability inventory: Uses cat, ls, find, and git for data gathering (SKILL.md).
Sanitization: None detected, but the data is consumed for fingerprinting, not shell execution.

project-fingerprinting