The Agent Skills Directory

[COMMAND_EXECUTION]: The scripts/clone_for_analysis.py script executes shell commands using subprocess.run to perform git clone operations. While it uses a list for arguments to mitigate basic shell injection, it processes URLs derived from an external JSON file.
[EXTERNAL_DOWNLOADS]: The skill fetches a community addon database from a public GitHub repository (cqb13/meteor-addon-scanner). This database contains metadata and repository links used by the skill's scripts.
[INDIRECT_PROMPT_INJECTION]: The skill creates an attack surface by ingesting untrusted data from an external repository.
Ingestion points: The scripts/filter_addons.py script downloads addons.json from a third-party GitHub account.
Boundary markers: None. The agent is instructed to use the filtered output for reference and analysis without specific delimiters or warnings about embedded instructions.
Capability inventory: The skill has the capability to write files and execute network operations through git clone via scripts/clone_for_analysis.py.
Sanitization: No sanitization is performed on the metadata fields (names, descriptions) which the agent may read into its context.

meteor-addon