The Agent Skills Directory

[PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection through untrusted web content. Ingestion points: scripts/extract_url.py fetches data from external URLs; the pipe mode processes stdin or local files. Boundary markers: While scripts/extract_url.py wraps output in a JSON envelope, it does not provide explicit boundary markers or instructions for the agent to ignore instructions embedded within the extracted text. Capability inventory: The skill has Bash and Write permissions, enabling potential misuse if the agent follows malicious instructions found in scraped content. Sanitization: No sanitization, validation, or filtering is performed on the article text before it is returned to the agent.
[COMMAND_EXECUTION]: The skill safely executes the trafilatura CLI binary via Python scripts. Evidence: Found in scripts/doctor.py, scripts/extract_url.py, scripts/list_urls.py, and scripts/check.py. Safety: All subprocess.run calls utilize list-based arguments rather than shell strings, preventing command injection vulnerabilities.

trafilatura