The Agent Skills Directory

[EXTERNAL_DOWNLOADS] (LOW): The skill is installed via npx skills add bryantleft/bruhs-skills and uses npx for the Linear MCP server. These are necessary for the skill's stated purpose but involve fetching external resources.
[REMOTE_CODE_EXECUTION] (LOW): The /bruhs spawn command uses pnpm create to scaffold projects, which downloads and executes remote templates. This is standard for development tools but constitutes runtime remote code execution.
[COMMAND_EXECUTION] (LOW): The skill performs extensive CLI operations using git and gh, including branch deletion and PR management. The risk is mitigated by the use of AskUserQuestion to require human approval for destructive or significant actions.
[PROMPT_INJECTION] (LOW): The /bruhs peep command fetches pull request comments from GitHub using gh api. This creates a surface for Indirect Prompt Injection (Category 8). Evidence Chain: 1. Ingestion points: GitHub PR comments (external data). 2. Boundary markers: No explicit delimiters or warnings used when processing these comments. 3. Capability inventory: The agent has access to the filesystem, Git, GitHub CLI, and NPX. 4. Sanitization: No sanitization of comment content is documented before the agent processes them.

bruhs