The Agent Skills Directory

[COMMAND_EXECUTION]: The skill facilitates the execution of local scripts and development tools using bun and node. This includes linting (bun run lint), end-to-end testing (bun run e2e), and a custom record auditing script (scripts/audit-records.mjs). These commands are standard for project validation but involve execution of local scripts.
[EXTERNAL_DOWNLOADS]: The skill instructs the agent to manage and install the @temps-sdk/ds package using bun add. This is a standard package management operation for a vendor-owned resource.
[INDIRECT_PROMPT_INJECTION]: The skill reads from various documentation and source files to determine design rules, which could theoretically allow malicious content within those files to influence agent behavior.
Ingestion points: Documentation files such as RULES.md, brand-guidelines.md, and design-system-handoff.md, along with React component files in design-system/src/sections/.
Boundary markers: No explicit delimiters or boundary markers are defined to isolate untrusted content from the agent's primary instructions.
Capability inventory: The skill allows for command execution via bun and node for linting and auditing tasks.
Sanitization: There is no evidence of sanitization or validation of the file content before it is processed by the agent.

temps-design-system