The Agent Skills Directory

[PROMPT_INJECTION]: The skill exhibits a vulnerability surface for indirect prompt injection by ingesting untrusted data from the web.
Ingestion points: The competitor-scraper.py script retrieves content from public competitor URLs which is then processed by the agent.
Boundary markers: The instructions do not define explicit delimiters or instructions to ignore embedded commands within the scraped content.
Capability inventory: The skill has capabilities to read local filesystem data (~/.claude-marketing/), execute internal scripts, and generate schema markup.
Sanitization: While the process mentions "Accuracy verification" and a "brand-guardian" agent, there is no evidence of technical sanitization or filtering of the raw external content before it is interpolated into the agent's context.
[COMMAND_EXECUTION]: The skill explicitly references and utilizes several Python scripts (schema-generator.py, competitor-scraper.py, content-scorer.py). These scripts represent the primary functional logic but also define the skill's execution boundary.
[DATA_EXFILTRATION]: The skill accesses local configuration and profile files located at ~/.claude-marketing/brands/. While these appear to be application-specific files for brand management, the access to absolute paths outside of the immediate project directory is a designated behavior for data exposure analysis.

competitor-pages