The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill ingests data from local project files (package.json, components.json) using scripts/audit-project.mjs to detect the project stack and provide guidance. * Ingestion points: The skill reads configuration files from the current working directory to identify installed dependencies and framework settings. * Boundary markers: The scripts do not use explicit delimiters or warnings when presenting the content of these files to the agent, which could allow malicious content in a project file to influence the agent's behavior. * Capability inventory: The skill has the ability to execute internal Node.js scripts and write files to the filesystem (.agents/frontend/ and registry/ directories). * Sanitization: No sanitization or filtering is performed on the data read from the local JSON files before it is processed.
[COMMAND_EXECUTION]: The skill is configured to execute its own internal Node.js scripts to perform project audits and scaffolding. * Evidence: The allowed-tools configuration in SKILL.md restricts execution to node ${AGENTS_SKILLS_DIR}/scripts/*. * The provided scripts (audit-project.mjs, init-house-registry.mjs, plan-surface.mjs, scaffold.mjs) use standard Node.js APIs for filesystem management and do not exhibit malicious behavior such as privilege escalation or network exfiltration.

product-ui-kit