The Agent Skills Directory

[COMMAND_EXECUTION]: The skill utilizes shell commands including gh, git, jq, sed, and grep to interact with GitHub repositories and process workflow data. These operations target the vendor's own repository (nubjs/nub) and are appropriate for CI diagnostic tasks.
[INDIRECT_PROMPT_INJECTION]: The skill ingests untrusted data from external CI logs, creating a potential surface for indirect prompt injection.
Ingestion points: GitHub Actions failure logs retrieved via gh run view --log-failed and check-run metadata from the GitHub API.
Boundary markers: Absent. There are no instructions or delimiters to help the agent distinguish between log content and operational commands.
Capability inventory: The skill has shell execution capabilities via gh, git, sed, jq, and grep.
Sanitization: The skill strips ANSI escape sequences from logs using sed for readability, but does not sanitize or filter the content for malicious prompt instructions.

ci-triage