The Agent Skills Directory

[COMMAND_EXECUTION]: The skill is designed to interact with the host system via the shell tool using the nb api data-modeling CLI. This tool is used to perform all database modeling operations including creating, updating, and destroying collections and fields.
Evidence: SKILL.md defines the shell tool in its allowed-tools frontmatter and specifies a range of CLI operations under the 'Final Command Surface' section.
[INDIRECT_PROMPT_INJECTION]: The skill is vulnerable to indirect prompt injection because it ingests untrusted user requirements to generate and execute shell commands. There are no explicit instructions for the agent to sanitize user-provided identifiers (like collection or field names) or expressions (like formula strings) before interpolating them into shell commands.
Ingestion points: The skill processes user-supplied modeling requirements in SKILL.md.
Boundary markers: No delimiters or "ignore embedded instructions" warnings are specified for the user-provided data.
Capability inventory: The agent has shell access to execute database schema mutations via the nb CLI.
Sanitization: No sanitization, escaping, or validation logic is defined for user-provided strings used as command arguments.
[EXTERNAL_DOWNLOADS]: The skill references external resources including documentation and plugin packages related to the NocoBase ecosystem.
Evidence: references/fields/plugins/formula.md contains links to docs.nocobase.com. references/plugin-provided-capabilities.md lists various @nocobase/ scoped NPM packages as dependencies for advanced field types. These are official vendor resources.

nocobase-data-modeling