The Agent Skills Directory

Command Execution: The skill performs several local operations using cargo, git, and gh (GitHub CLI). These are standard tools for Rust development and version control; however, they involve the execution of scripts and commands within the project environment.
Indirect Prompt Injection: The workflow processes commit messages and Pull Request metadata to generate release notes. Since this data originates from repository history and external contributors, it could potentially contain adversarial instructions that might influence the agent's behavior.
Ingestion points: Commit logs retrieved via git log and PR metadata retrieved via the gh tool (SKILL.md steps 6 and 7).
Boundary markers: The skill does not currently specify the use of delimiters or instructions to ignore embedded commands within the ingested data.
Capability inventory: The agent has the ability to execute shell commands and interact with the repository via cargo and git.
Sanitization: The instructions do not include specific steps to sanitize or filter the content of commit messages before processing.
Dynamic Execution: The skill utilizes cargo xtask, a pattern that involves compiling and running project-specific automation code. This is a standard practice in the Rust ecosystem for build and release tasks, representing the execution of logic defined within the local repository.

bump-version