The Agent Skills Directory

[SAFE]: The skill incorporates a proactive 'Secret Leak Scan' (Task 4) using Grep to detect sensitive patterns, including bearer tokens and configuration file references, preventing accidental inclusion of secrets in releases.
[SAFE]: Adheres to the principle of least privilege by preparing but not executing high-risk commands such as git push, explicitly instructing the agent to provide these commands for the user to run manually.
[PROMPT_INJECTION]: Indirect prompt injection surface identified in the changelog generation process where the skill processes untrusted commit messages via git log. Ingestion points: commit history retrieved via git log. Boundary markers: absent. Capability inventory: file system modification (Write, Edit) and shell execution (Bash). Sanitization: absent. Note: This surface is inherent to the skill's primary function.

releasing