The Agent Skills Directory

[PROMPT_INJECTION]: The skill is susceptible to Indirect Prompt Injection because it ingests untrusted data from GitHub issue and PR titles/bodies to generate suggestions.
Ingestion points: External content is fetched via gh issue view and gh pr view in SKILL.md.
Boundary markers: There are no explicit instruction-delimiters used when the agent processes the fetched content, though a logical human-in-the-loop checkpoint is required for execution.
Capability inventory: The skill has the capability to write to the repository via gh issue/pr edit (labels) and gh issue/pr comment (comments), and it appends logs to a local file system path.
Sanitization: No automated sanitization of the fetched GitHub content is performed; the skill relies entirely on the user to review suggestions before approval.
[COMMAND_EXECUTION]: The skill utilizes standard system tools and the GitHub CLI to perform its functions.
Uses cat to read local documentation (docs/triage-instructions.md).
Uses gh to interact with the NVIDIA/NemoClaw repository for both reading and writing.
Uses jq for processing JSON data returned by the GitHub API.
These commands are consistent with the stated purpose of the skill and the author's context.

nemoclaw-maintainer-triage