The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill processes untrusted data from repository history, including commit messages, pull request titles, and changelog contents, to validate releases and generate reports.
Ingestion points: The skill reads commit messages (git log), pull request metadata (gh pr list), and file differences (gh pr diff) as described in Step 4 of the workflow.
Boundary markers: The skill implements a "gather facts" workflow and explicitly mandates that the "human decides" before any merge actions, providing a human-in-the-loop safeguard.
Capability inventory: The skill has the capability to create pull requests (gh pr create) and merge them (gh pr merge) into protected branches.
Sanitization: Instructions in the reference documentation specify resolving configuration data into shell variables using specific jq patterns to prevent unsafe interpolation into command flags.
[DYNAMIC_EXECUTION]: The skill utilizes a local shell script (templates/resolve-config.sh) to resolve and merge configuration profiles from .tituskirch-skills.json at runtime.

release