The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill analyzes the project's source code and configuration files to build the dashboard, which presents a surface for indirect prompt injection if the local files are maliciously crafted. ● Ingestion points: The scripts/inventory-data-model.mjs script reads Prisma schemas, SQL migrations, Drizzle/Mongoose definitions, and package.json from the local filesystem (e.g., SKILL.md step 2). ● Boundary markers: There are no explicit delimiters or specific instructions for the agent to disregard instructions embedded within the analyzed project files. ● Capability inventory: The skill executes a bundled script to read the filesystem and is designed to create or modify code files to implement the dashboard UI. ● Sanitization: The skill does not explicitly describe sanitization or filtering of names and metadata extracted from the project files before they are used to influence the agent's code generation tasks.
[COMMAND_EXECUTION]: The skill uses node to run a bundled inventory script (scripts/inventory-data-model.mjs). This script performs a static analysis of the local project directory using standard node:fs APIs to identify data models and the tech stack. It does not perform network operations or execute arbitrary shell commands beyond the localized discovery script.

admin-dashboard