The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The script scripts/conversion_audit.py uses urllib.request.urlopen to fetch content from arbitrary external URLs provided by the user via the --url argument, which allows for outbound network requests.
[INDIRECT_PROMPT_INJECTION]: The skill processes untrusted HTML content from external websites or local files to provide CRO recommendations, creating a vulnerability where malicious instructions embedded in the HTML could attempt to manipulate the agent.
Ingestion points: scripts/conversion_audit.py reads data from remote URLs or local file paths.
Boundary markers: The instructions do not specify any delimiters or warnings to the agent to disregard instructions within the processed data.
Capability inventory: The skill uses urllib.request for network access and file reading tools to ingest data.
Sanitization: There is no evidence of HTML sanitization or filtering to remove potential prompt injection payloads before the agent processes the audit results.
[METADATA_POISONING]: The SKILL.md frontmatter contains a future date (2026-03-06) in the updated field, which is misleading regarding the skill's actual creation or update history.
[COMMAND_EXECUTION]: The skill involves the execution of a Python script (scripts/conversion_audit.py) to perform its audit functions, which is a standard capability for this type of skill.

page-cro