The Agent Skills Directory

[PROMPT_INJECTION]: The skill processes untrusted external data (such as web search results, LinkedIn activity, and blog posts) to synthesize sales strategies and create outreach sequences. This creates a surface for indirect prompt injection, where malicious instructions embedded in researched content could influence the agent's actions.
Ingestion points: Web search results (news, social media, articles), LinkedIn profiles, and CRM/Amplemarket data records as described in Phase 1 of the SKILL.md.
Boundary markers: There are no explicit instructions to the agent to treat research results as untrusted data or to use delimiters for external content to prevent command following.
Capability inventory: The skill possesses write capabilities through tools like create_sequence and add_leads_to_sequence, along with investigative capabilities like search_people, enrich_person, and get_messaging_settings in SKILL.md.
Sanitization: The instructions do not define sanitization or validation logic for data retrieved from the web before it is incorporated into outreach messaging or sequence generation.

torpedo