The Agent Skills Directory

[COMMAND_EXECUTION]: The skill uses sudo in SKILL.md to perform system-level package installations and keyring management, which requires administrative access.
[COMMAND_EXECUTION]: The helper script scripts/project-helpers.sh employs the eval command within its execute_with_retry function to execute dynamically constructed shell commands, a pattern that can lead to command injection if input parameters are not strictly sanitized.
[REMOTE_CODE_EXECUTION]: The functions ensure_gh_cli and update_item_status in scripts/project-helpers.sh attempt to execute local shell scripts (ensure-gh-cli.sh, ensure-dependencies.sh, and graphql-queries.sh) that are not present in the skill package, preventing a full audit of the execution flow.
[EXTERNAL_DOWNLOADS]: The skill performs downloads from cli.github.com for tool setup; this source is recognized as a well-known technology service.
[PROMPT_INJECTION]: The skill is vulnerable to indirect prompt injection because it ingests untrusted data from GitHub issue and pull request titles and states in scripts/graphql-queries.sh and scripts/project-helpers.sh. There are no boundary markers or explicit sanitization steps to prevent malicious content in these fields from influencing agent behavior.

managing-projects