The Agent Skills Directory

[COMMAND_EXECUTION]: The skill constructs and executes shell commands in SKILL.md (Phase 2) by directly interpolating values from the client's configuration file (clients/<client>/config/industry-scanner.json). Parameters such as blog_urls, subreddits, reddit_keywords, twitter_query, hn_query, and review_url are passed as arguments to python3 scripts. If these configuration fields contain shell metacharacters or maliciously crafted strings, it could lead to arbitrary command execution on the host system.
[PROMPT_INJECTION]: The skill is vulnerable to indirect prompt injection because it ingests large volumes of untrusted data from various external platforms to generate strategic business advice and tactics.
Ingestion points: Phase 2 of SKILL.md documents data collection from web searches, industry blogs, Reddit, Twitter/X, LinkedIn, Hacker News, RSS feeds, newsletter inboxes, and review sites (G2, Capterra, Trustpilot).
Boundary markers: None. The instructions do not specify the use of delimiters or 'ignore' instructions for the fetched content during the categorization and strategy generation phases.
Capability inventory: The agent can execute shell commands to trigger scrapers (Phase 2) and write report files to the local file system (Phase 5).
Sanitization: None. There is no evidence of sanitization, filtering, or validation of the external content before it is processed by the LLM for relevance rating and opportunity generation.

industry-scanner