The Agent Skills Directory

[PROMPT_INJECTION]: The skill is vulnerable to indirect prompt injection (Category 8) because it ingests untrusted data from the internet.
Ingestion points: Competitor websites are scraped via Playwright's snapshot tool in the sub-agent instructions.
Boundary markers: Absent. There are no delimiters or instructions to ignore malicious commands embedded in the scraped HTML or text content.
Capability inventory: The main agent has access to Bash, Write, and Agent tools, while sub-agents have the ability to write JSON reports to the file system.
Sanitization: No sanitization or validation is applied to the content retrieved from external sites before it is passed to the LLM for extraction.
[COMMAND_EXECUTION]: The skill uses the Bash tool to execute several local Python scripts for data merging, CSV-to-XLSX conversion, diff calculation, and summary generation.
[EXTERNAL_DOWNLOADS]: The skill's execution algorithm includes a step to install the Playwright MCP tool using npx @playwright/mcp@latest from the public npm registry if the tool is missing. While this targets a well-known service, it involves executing third-party code at runtime.

competitive-analysis