The Agent Skills Directory

[SAFE]: The skill provides a structured methodology for business strategy auditing, including AI-era resilience and classical moat analysis.
[DATA_EXPOSURE]: The skill accesses local project-related files (e.g., .acumen/competitors.md, .acumen/features.md) to gather context for its analysis. This is standard behavior for an agent skill working within a specific project directory, and the skill contains no mechanisms to exfiltrate this data externally.
[PROMPT_INJECTION]: The skill uses directive language such as "MANDATORY PREPARATION" and "non-negotiable" to define its operational protocol. These instructions are within the scope of defining the agent's task-specific persona and output quality, rather than attempting to bypass safety filters or extract system prompts.

defensibility