The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill processes untrusted data from external social media profiles, creating a surface for indirect prompt injection attacks.
Ingestion points: The skill fetches raw data using mcp__claude-in-chrome__get_page_text and mcp__claude-in-chrome__read_page to collect tweet text, engagement metrics, and profile information.
Boundary markers: There are no instructions for the agent to treat the ingested tweet content as data rather than instructions, nor are there delimiters to wrap the external text.
Capability inventory: The agent utilizes browser automation tools including mcp__claude-in-chrome__navigate and mcp__claude-in-chrome__javascript_tool which could be manipulated if the agent follows instructions found within a tweet.
Sanitization: The workflow does not include any steps to sanitize or filter the scraped text to remove potential command-like phrases before the agent processes the information.

tweet-series-extractor