The Agent Skills Directory

[DATA_EXPOSURE]: The skill reads from local files such as opc-doc/outputs/01-resource-audit/inventory.md and opc-doc/inputs/market-notes.md. These operations are localized to the project directory and are required for the skill's business logic.
[COMMAND_EXECUTION]: There are no shell commands, privilege escalation attempts (sudo), or unsafe script executions within the instruction files.
[REMOTE_CODE_EXECUTION]: The skill does not perform any network operations, download external scripts, or install third-party packages.
[INDIRECT_PROMPT_INJECTION]: The skill processes untrusted user input from market-notes.md and raw-ideas.md. While it lacks explicit boundary markers or sanitization, its capabilities are restricted to generating and writing text files, which minimizes the risk of significant harm.

opc-niche-positioning