The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill ingests data from external lead profiles which may contain adversarial content intended to influence message generation.
Ingestion points: Data enters the agent context via the find_people tool, which returns lead names, titles, companies, and LinkedIn URLs.
Boundary markers: The instructions do not define specific delimiters or instructions to treat lead data as untrusted text during message generation.
Capability inventory: The skill uses save_sequence_draft to create outreach sequences and relies on the write-cold-message skill for content creation.
Sanitization: There is no mention of sanitizing or filtering lead-provided data before it is used in prompt interpolation for sequence drafting.
Mitigation: A significant safeguard is present as the skill explicitly states that nothing is sent until the user approves the sequence, and the user is shown the lead details during the review step.

find-buyers