The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill collection includes multiple tools designed to ingest and analyze untrusted external data, creating a surface for indirect prompt injection attacks.
Ingestion points: Tools like customer_interview_analyzer.py, extract_citations.py, and codebase_analyzer.py read content from external files (transcripts, research papers, and codebases) and provide them to the agent for analysis.
Boundary markers: The instructions lack explicit delimiters or "ignore embedded instructions" warnings for the agent when processing this untrusted content.
Capability inventory: The skill possesses powerful capabilities that could be abused if an injection is successful, including executing shell commands (git via changelog_generator.py) and writing files to the local filesystem (project_bootstrapper.py and prd_scaffolder.py).
Sanitization: Input data undergoes minimal processing, and no sanitization is performed to filter out malicious language instructions.
[COMMAND_EXECUTION]: The changelog_generator.py script executes the git log command using the subprocess module to generate release notes. While the script avoids using a shell and provides a timeout, the commit_range parameter is constructed from user-supplied or agent-supplied arguments, which could lead to command argument manipulation if not handled carefully.
[EXTERNAL_DOWNLOADS]: The documentation provides instructions for installing the skill and its dependencies using git clone, npm, and the clawhub registry. These point to the author's official GitHub repository and standard software registries.

product-skills