The Agent Skills Directory

[PROMPT_INJECTION]: The skill is vulnerable to indirect prompt injection due to its ingestion of large volumes of untrusted data from external sources.
Ingestion points: Data enters the agent's context from web search results, public blogs, Reddit threads, Twitter feeds, LinkedIn posts, Hacker News, and newsletters (Phase 2).
Boundary markers: There are no defined delimiters or specific instructions within the skill logic to isolate or ignore potentially malicious instructions embedded in the external content.
Capability inventory: The skill has access to local client configuration and context files, can write intelligence reports to the filesystem, and executes subprocesses for data collection.
Sanitization: The skill lacks mechanisms to sanitize or filter the content collected from external sources before it is processed by the language model for strategy generation.
[COMMAND_EXECUTION]: The skill executes multiple local Python scripts via shell commands as part of its core orchestration logic.
Evidence: Phase 2 explicitly lists bash commands such as python3 skills/blog-scraper/scripts/scrape_blogs.py and python3 skills/reddit-scraper/scripts/search_reddit.py which are called with arguments derived from configuration files and lookback periods.

industry-scanner