The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and process several external project files (e.g., ТЗ.md, DESIGN.md, голос.md) which could contain malicious instructions or unexpected content intended to manipulate agent behavior.
Ingestion points: The skill reads documentation from the spec/, design/, and content/ directories within the project.
Boundary markers: There are no instructions provided to wrap the ingested content in delimiters or to ignore potential instructions embedded within those files.
Capability inventory: The agent is authorized to perform filesystem operations (creating .env files), network operations (cloning repositories, connecting to APIs), and package installations.
Sanitization: No mechanisms for validating, escaping, or filtering the content of the external files are defined.

build-architect