The Agent Skills Directory

[SAFE]: The skill performs standard file operations (reading and writing) within a local project directory (founder/) to process business data, which is consistent with its stated purpose of offer generation.
[DATA_EXPOSURE]: The skill accesses local files containing potentially sensitive business information such as founder/numbers.json (costs) and founder/board.md (board conditions). However, this access is localized for processing, and there are no indicators of external data exfiltration.
[INDIRECT_PROMPT_INJECTION]: The skill ingests data from external sources such as founder/panel/results.md (buyer feedback). While this constitutes an attack surface for indirect prompt injection, the skill lacks high-risk execution sinks like remote network requests or dynamic code evaluation, and its output is confined to local documentation updates, keeping the risk level at safe/negligible.

founder-offer