The Agent Skills Directory

[DATA_EXPOSURE_AND_EXFILTRATION]: The skill provides instructions for managing sensitive authentication tokens such as GITLAB_TOKEN and CI_JOB_TOKEN within development and CI/CD environments. It emphasizes best practices by explicitly warning against printing these tokens and detailing the correct header mapping for the GitLab Releases API to prevent credential leakage or authorization failures.
[INDIRECT_PROMPT_INJECTION]: The skill enables the agent to ingest external data from GitLab repositories, including release notes (Markdown) and structured release lists (JSON).
Ingestion points: CHANGELOG.md files, standard output from glab release list, and release descriptions from the platform.
Boundary markers: None explicitly defined in the instructions.
Capability inventory: Execution of glab CLI subcommands (create, delete, download, etc.).
Sanitization: No specific sanitization or filtering of ingested data is described. While the ingestion of untrusted repository content represents a potential attack surface, the skill does not provide methods for this content to bypass security filters or influence the execution of high-risk commands.
[COMMAND_EXECUTION]: The skill instructs the agent on the use of the glab command-line utility. All provided examples and subcommand descriptions are consistent with official GitLab documentation and do not include obfuscated commands, unsafe shell piping, or execution of arbitrary scripts from untrusted sources.

glab-release