The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and process data from external sources, creating a surface for potential indirect prompt injection attacks.
Ingestion points: The agent is instructed to read local files in 'node_modules/@mastra/' (referenced in SKILL.md and embedded-docs.md) and fetch remote documentation from 'https://mastra.ai/llms.txt' (referenced in remote-docs.md).
Boundary markers: No explicit delimiters or instructions are provided to the agent to sanitize or ignore potential instructions embedded within these external documents.
Capability inventory: The skill provides the agent with file read capabilities (grep, cat) and execution capabilities (npm install, npx, npm run).
Sanitization: There is no evidence of input validation or content filtering for the data retrieved from remote URLs or local node_modules.
[COMMAND_EXECUTION]: The skill contains instructions for the agent to execute various shell commands to inspect the local environment and project structure.
Evidence: Commands include 'ls node_modules/@mastra/', 'grep -r', and 'cat' to read source maps and type definitions in 'SKILL.md' and 'embedded-docs.md'. It also includes 'npm run dev' to start a local development server.
[REMOTE_CODE_EXECUTION]: The skill encourages the use of remote package managers and execution tools to manage the Mastra project.
Evidence: Instructions include 'npm create mastra@latest', 'npm install @mastra/core@latest', and 'npx @mastra/codemod@latest' in 'create-mastra.md' and 'migration-guide.md'. These commands download and execute code from the framework's official package registry.
[EXTERNAL_DOWNLOADS]: The skill documentation frequently references and instructs the agent to fetch content from external URLs.
Evidence: Fetches documentation and project guidelines from 'https://mastra.ai/llms.txt' and other paths under the 'mastra.ai' domain using a fetch tool.

mastra