The Agent Skills Directory

[COMMAND_EXECUTION]: The skill makes extensive use of the GitHub CLI (gh) and jq to query and modify repository data, including issue creation and project board updates.
[INDIRECT_PROMPT_INJECTION]: The skill processes data from external sources (GitHub issues and project boards). This creates an attack surface where malicious content embedded in existing issues could potentially influence the agent's actions during the verification or creation process.
Ingestion points: Fetches issue details (gh issue view) and project items (gh project item-list) and incorporates them into the workflow.
Boundary markers: None present in the instructions to prevent the agent from following instructions embedded in issue content.
Capability inventory: Shell execution for GitHub CLI commands (create, view, edit) and JSON parsing via jq.
Sanitization: The instructions do not specify sanitization or escaping for user-provided strings used in shell commands.
[DYNAMIC_EXECUTION]: Shell scripts are constructed dynamically by interpolating user input (such as Title, Description, and Priority) into gh command templates. This pattern is vulnerable to command injection if the agent does not properly escape these values before executing the resulting shell commands.

issue-prerequisite