The Agent Skills Directory

[REMOTE_CODE_EXECUTION]: The workflow includes a step to build a local package using npm ci && npm run build in the packages/skillcheck directory. This pattern involves downloading dependencies and executing arbitrary scripts defined in the repository, which is a common vector for supply chain attacks.
[COMMAND_EXECUTION]: The skill executes multiple shell commands to perform linting and validation, including node packages/skillcheck/bin/skillcheck.js, npx skillcheck, and npx agnix.
[EXTERNAL_DOWNLOADS]: The skill uses npm and npx to fetch and execute packages. While these are standard tools, the specific packages being invoked (skillcheck, agnix) are not from recognized trusted organizations or well-known services defined in the security policy.
[INDIRECT_PROMPT_INJECTION]: The skill ingests untrusted data by reading rubric files (*/references/*rubric*.md) and integrating them into agent prompts for review and alignment tasks.
Ingestion points: Rubric files discovered during the workflow in SKILL.md and review-prompt.md.
Boundary markers: None identified. The skill does not use specific delimiters or instructions to ignore potential commands within the rubric content.
Capability inventory: The skill has significant capabilities including file writing, network access (via npm/npx), and arbitrary command execution.
Sanitization: No sanitization or validation of the ingested markdown content is performed before it is processed by the agent or subagents.

writing-rubrics