The Agent Skills Directory

[SAFE]: The skill implements a restricted operating contract that limits modifications to local version-related files and explicitly forbids creating tags, commits, or publishing releases. It provides specific command guardrails, such as using the --no-git-tag-version flag for npm, to ensure consistent and safe behavior.
[PROMPT_INJECTION]: The skill possesses a surface for indirect prompt injection as it derives version recommendations from commit messages and changelog content which could contain adversarial instructions. 1. Ingestion points: Git commit history and CHANGELOG.md unreleased sections. 2. Boundary markers: The skill does not define specific delimiters or isolation protocols for the processed text content. 3. Capability inventory: The skill can modify local files and execute package manager CLI tools as documented in references/commands.md. 4. Sanitization: No explicit sanitization or filtering of commit/changelog data is performed before reasoning.

version-bump