The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill's core methodology involves using browser automation (Playwright) to ingest data from various external industry news and market data sources. This creates a surface for indirect prompt injection if an attacker-controlled website contains instructions intended to manipulate the agent's behavior during data extraction.
Ingestion points: Browser automation workflow using Playwright to navigate to external URLs in SKILL.md.
Boundary markers: None detected; the skill does not explicitly instruct the agent to ignore or delimit instructions found within the processed web content.
Capability inventory: Access to Bash, Write, and mcp__playwright_* tools provides a broad execution environment.
Sanitization: None detected; the skill relies on visual analysis of screenshots but lacks content filtering for text extraction.
[DYNAMIC_EXECUTION]: The output structure defined in SKILL.md includes a scripts/ directory for "Generated research code." The generation of executable code at runtime based on research tasks introduces a dynamic execution surface, although it appears to be a standard part of the research workflow template.

industry-research