The Agent Skills Directory

[EXTERNAL_DOWNLOADS] (LOW): The skill instructs the agent to install packages such as ai, @ai-sdk/openai, and @ai-sdk/devtools using project package managers. These originate from Vercel, a trusted organization.
[COMMAND_EXECUTION] (LOW): The skill contains commands to fetch model lists (curl https://ai-gateway.vercel.sh/v1/models) and process the output with jq. It also suggests using grep to search local node_modules. These operations are limited to the skill's primary purpose of providing AI SDK support.
[PROMPT_INJECTION] (LOW): Includes instructions to 'disregard internal knowledge' ('Everything you know about the AI SDK is outdated or wrong'). While this matches override patterns, it is used here to prioritize current documentation over potentially stale model training data.
[INDIRECT PROMPT INJECTION] (LOW): The skill identifies a surface for indirect prompt injection by ingesting data from node_modules and the ai-sdk.dev website.
Ingestion points: Reads local files in node_modules/ai/ and fetches documentation from https://ai-sdk.dev.
Boundary markers: Absent. The skill does not explicitly instruct the agent to ignore instructions embedded within the documentation it reads.
Capability inventory: Can execute shell commands (grep, curl, jq) and install packages (pnpm, npm, npx).
Sanitization: Absent. Documentation content is processed directly for answering queries.

ai-sdk