The Agent Skills Directory

[COMMAND_EXECUTION]: The skill utilizes common developer tools including git and rg (ripgrep) to fetch branch metadata, read project files, and identify changes in the repository. These operations are restricted to the local repository context and the configured remote.
[DATA_EXPOSURE_AND_EXFILTRATION]: Access is limited to project-specific files (packages/core/src/version.ts and packages/docs). There is no evidence of unauthorized access to sensitive system credentials or network exfiltration to external domains.
[INDIRECT_PROMPT_INJECTION]: The skill ingests data from the repository via git show and git diff. While this constitutes an attack surface where an attacker with write access to the repository could attempt to insert instructions into the version file or documentation, the skill's logic is highly specific to version incrementing, which limits the potential for instruction diversion.
[METADATA_POISONING]: The skill's metadata accurately reflects its behavior and project scope (Remotion), with no deceptive or misleading information provided.

update-version