The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill exhibits an attack surface for indirect prompt injection by processing untrusted data from external sources. * Ingestion points: Project structure, configuration files, dependencies, and build error logs (as defined in the 'Behavioral Flow' and 'Error Handling' sections). * Boundary markers: The skill does not define any boundary markers or delimiters to isolate untrusted project data from the agent's primary instructions. * Capability inventory: The skill utilizes powerful tools including shell execution ('Bash'), file reading ('Read'), and file writing ('Write') as specified in the 'Tool Coordination' section. * Sanitization: No evidence of input validation, sanitization, or escaping of data extracted from project files or logs before it is used in the build process or reported to the user.

sc-build