The Agent Skills Directory

[PROMPT_INJECTION]: The skill contains explicit instructions for the agent to override default system behaviors related to AI attribution markers, such as "Co-Authored-By" trailers. These instructions are intended to ensure the agent's output complies with human-centric project standards and do not target safety or security guardrails.
[COMMAND_EXECUTION]: The skill facilitates the execution of shell commands using git and glab. It processes external inputs, such as Jira task titles, to generate parameters for branch creation. The skill includes specific sanitization rules (e.g., kebab-case conversion, character stripping, and length limits) to transform raw titles into safe branch names, effectively mitigating potential command injection risks through these inputs.
[SAFE]: The skill demonstrates security best practices by explicitly instructing the agent to avoid committing sensitive information, specifically naming files like .env, credentials.*, *.key, and *.pem as items that must not be included in commits.

gitlab-flow