The Agent Skills Directory

[SAFE]: The skill provides instructions for a behavior-preserving refactor of Mastra agents into a file-based convention. The mapping of constructor options to specific files (config.ts, instructions.md, tools/*.ts) follows the framework's official conventions.
[COMMAND_EXECUTION]: The skill recommends using npx mastra dev and npx mastra build to verify the migration. These commands are part of the Mastra CLI and are necessary for the discovery and validation of file-based agents as intended by the framework.
[PROMPT_INJECTION]: The skill processes user-supplied source code files and interpolates instructions into new agent files. This creates an indirect prompt injection surface where instructions within the processed code could potentially influence the agent during the migration task. This is documented as a low-risk inherent surface for code-editing skills.
Ingestion points: The skill reads agent definitions from source files such as src/mastra/index.ts and src/mastra/agents/*.ts.
Boundary markers: The skill does not instruct the agent to use specific delimiters or warnings when processing or migrating instruction text.
Capability inventory: The skill involves file creation and modification, alongside execution of the Mastra CLI tools.
Sanitization: None present; the skill performs a direct transcription of existing agent logic to new file structures.

mastra-file-agents