The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill ingests and processes content from external GitLab wiki pages, creating a surface for indirect prompt injection where untrusted data could influence agent actions.
Ingestion points: Multiple workflows in SKILL.md (e.g., Workflow 3 and Workflow 4) demonstrate the retrieval of wiki page content using glab api and jq (e.g., content=$(glab api ... | jq -r '.content')).
Boundary markers: The skill lacks explicit boundary markers or instructions to the agent to ignore potentially malicious instructions embedded within the retrieved wiki text.
Capability inventory: The skill is granted Bash tool access, allowing it to execute system commands, process retrieved data with jq, and perform file system operations like creating backups.
Sanitization: While the provided bash snippets use double-quoting for variables (e.g., "$content"), the skill does not implement specific sanitization or filtering for the content retrieved from GitLab before it is used in subsequent operations or processed by the agent.
[COMMAND_EXECUTION]: The skill relies on the Bash tool to perform its primary functions via the glab CLI.
Evidence: The skill defines several command patterns and complex workflows that use bash loops, pipes, and command substitution to automate tasks like backing up all wiki pages or migrating content between projects.

gitlab-wiki