The Agent Skills Directory

[PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection due to its core functionality of scraping and analyzing untrusted web content.
Ingestion points: Web search queries and full URL content extraction are performed via the Rube MCP integration (SKILL.md, Phase 3).
Boundary markers: The skill lacks explicit instructions or delimiter-based markers to prevent the model from obeying instructions embedded within the retrieved web data.
Capability inventory: The agent has access to highly capable tools including mcp__rube__RUBE_REMOTE_BASH_TOOL, Bash, and Write for file system operations (SKILL.md, Tool Coordination).
Sanitization: There is no evidence of sanitization or safety filtering of the raw findings before they are passed to the PAL ThinkDeep analysis phase.
[COMMAND_EXECUTION]: The skill incorporates tools capable of executing shell commands as part of its standard operational flow.
Evidence: The documentation specifies the use of the local Bash tool and a vendor-specific mcp__rube__RUBE_REMOTE_BASH_TOOL intended for processing large volumes of data (SKILL.md, MCP Integration).

sc-research