The Agent Skills Directory

[PROMPT_INJECTION]: The skill uses repetitive, high-urgency phrases like 'MANDATORY IMPORTANT MUST ATTENTION' and 'NO EXCEPTIONS' to strictly enforce workflow steps. This style is used to constrain agent autonomy and prioritize skill-specific instructions over general safety or operational guidelines.
[PROMPT_INJECTION]: The skill presents an indirect prompt injection vulnerability surface. Ingestion points: It ingests untrusted data from user inputs and 'WebSearch/WebFetch' tools during market analysis and idea capture. Boundary markers: No delimiters or isolation instructions are provided for external data to prevent the agent from interpreting it as instructions. Capability inventory: Access to 'Bash', 'Write', 'Edit', and 'TaskCreate' tools provides a high-impact surface should the agent process malicious payloads in external data. Sanitization: No validation or sanitization of external content is specified before the data is incorporated into business reports or execution plans.
[COMMAND_EXECUTION]: The skill frontmatter lists the 'Bash' tool as allowed, though the instructions do not explicitly describe its use, creating a potential for unexpected command execution during the business evaluation workflow.

business-evaluation