The Agent Skills Directory

[DATA_EXFILTRATION]: The skill reads potentially sensitive configuration and brand profile files from the local filesystem at ~/.claude-marketing/brands/_active-brand.json and ~/.claude-marketing/brands/{slug}/profile.json to apply brand voice and compliance rules.
[DYNAMIC_EXECUTION]: The skill relies on executing external Python scripts (schema-generator.py, competitor-scraper.py, and content-scorer.py) at runtime to perform its primary functions, including data extraction, quality scoring, and schema generation.
[INDIRECT_PROMPT_INJECTION]: The skill ingests untrusted data from external competitor websites via the competitor-scraper.py script. This data is then processed by the agent to create comparison content, creating a vulnerability surface where malicious instructions hidden on those websites could influence the agent's behavior.
Ingestion points: External competitor data retrieved by competitor-scraper.py from public sources.
Boundary markers: No explicit boundary markers or instructions to ignore embedded commands are present in the skill definition when processing scraped content.
Capability inventory: The skill has the ability to read local files (~/.claude-marketing/...) and execute shell commands/scripts (schema-generator.py, competitor-scraper.py, content-scorer.py).
Sanitization: The skill mentions "Accuracy verification" and utilizes a "brand-guardian" agent to check compliance, but lacks explicit technical sanitization of the scraped content to prevent prompt injection.
[COMMAND_EXECUTION]: The skill invokes shell commands to execute its associated Python scripts, specifically noted for schema-generator.py --type <Type>.

competitor-pages