The Agent Skills Directory

[SAFE]: No malicious patterns detected. The skill follows security best practices, including mandatory human-in-the-loop approval for all file modifications.
[COMMAND_EXECUTION]: The skill uses git log for codebase exploration and invokes internal utility scripts like effective_profile to process user preferences. These are standard operations within the agentera ecosystem.
[DATA_EXFILTRATION]: The skill reads local project files and user profiles to inform its vision generation. This data access is purposeful and scoped to the project environment, with no evidence of unauthorized data transmission.
[EXTERNAL_DOWNLOADS]: The skill performs web searches to gather domain context. This is a legitimate functional capability used to ground the vision in existing industry research and does not involve execution of untrusted code.

visionera