The Agent Skills Directory

[SAFE]: The skill's instructions focus on automating routine developer tasks and adhere to standard safety and security practices. No malicious code or obfuscation was detected.
[COMMAND_EXECUTION]: The skill uses git and gh (GitHub CLI) to manage branch lifecycles, including checking existing branches, fetching from remote repositories, and creating new branches from a clean state. All commands are standard and restricted to local repository management.
[PROMPT_INJECTION]: The skill ingests data from local documentation (found in docs/requirements/ or docs/plans/) and repository metadata (via gh pr list) to extract JIRA ticket IDs and identify base branches. While these are indirect prompt injection surfaces where external data could harbor malicious instructions, the risk is negligible. The extracted data is used exclusively to construct a branch name using a rigid lowercase kebab-case naming convention, which effectively sanitizes the input.
Ingestion points: Documentation files, user input, and PR metadata from GitHub.
Boundary markers: Absent.
Capability inventory: git branch, git checkout, git fetch, and gh repo view / gh pr list.
Sanitization: External data is constrained to a predefined naming template and forced into a kebab-case format.

create-branch