The Agent Skills Directory

[REMOTE_CODE_EXECUTION]: The 'Daily Skill Update Check' section specifies running npx -y skills@latest update -y. This command downloads and executes code from a public registry using an unpinned version and a generic name ('skills') that lacks a clear connection to the vendor 'ShiplightAI'. This creates a high-risk supply chain vulnerability.- [COMMAND_EXECUTION]: The skill frequently invokes shell commands such as npm install, npx playwright install, and npx shiplight test. These actions are performed based on instructions derived from local project files, which could be compromised.- [EXTERNAL_DOWNLOADS]: The skill performs automatic updates and dependency installations from the NPM registry and Playwright infrastructure.- [PROMPT_INJECTION]: The skill's multi-phase architecture is vulnerable to indirect prompt injection. 1. Ingestion points: Reads logic and context from test-context.md, test-spec.md, and test-plan.md. 2. Boundary markers: The skill does not use delimiters or instructions to separate ingested content from system instructions. 3. Capability inventory: The agent has the ability to execute shell commands, write files, and automate browsers. 4. Sanitization: There is no evidence of sanitization or validation of the input files before they influence the agent's executable actions.

create-tests