The Agent Skills Directory

[PROMPT_INJECTION] (LOW): Indirect prompt injection surface detected through GitLab data ingestion.
Ingestion points: The skill ingests untrusted data from the GitLab instance via glab milestone list and glab issue list commands (SKILL.md).
Boundary markers: Absent; data retrieved from external sources is not wrapped in protective delimiters or accompanied by instructions to ignore embedded commands.
Capability inventory: The skill uses Bash to execute glab commands and perform shell processing (SKILL.md).
Sanitization: Absent; the shell loop in Workflow 2 pipes output directly from a list command into an update command without validation or escaping.
[COMMAND_EXECUTION] (SAFE): The command usage is restricted to the glab binary for its intended administrative purpose. No arbitrary shell execution or unsafe user input interpolation into sensitive commands was detected.
[DATA_EXFILTRATION] (SAFE): No network exfiltration patterns or communication with non-whitelisted domains were identified. Data operations are confined to the user's GitLab environment.
[EXTERNAL_DOWNLOADS] (SAFE): No remote scripts or packages are downloaded or executed during the skill's runtime.

gitlab-milestone