The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill processes content from untrusted external sources, which could potentially contain instructions intended to influence the agent.
Ingestion points: Content is read from local README files, GitHub descriptions (via the gh CLI), and metadata from PyPI and npm registries.
Boundary markers: No explicit markers are used to delimit external content or instruct the agent to ignore embedded directives.
Capability inventory: The skill utilizes subprocess.run to call the gh CLI and urllib.request.urlopen for network communication in scripts/pitch.py.
Sanitization: The normalise function provides basic whitespace and case formatting but does not filter for potentially malicious semantic content.
[COMMAND_EXECUTION]: The script scripts/pitch.py executes the gh (GitHub CLI) tool to fetch repository information.
Evidence: The github_description function in scripts/pitch.py invokes subprocess.run with the gh command and user-provided repository arguments.

repo-pitch