The Agent Skills Directory

[COMMAND_EXECUTION]: The skill executes a local Python script located at scripts/mcp/morph_apply.py using the uv run command. This script is used to interact with the Morph Apply API and manage file system operations.- [DATA_EXFILTRATION]: As part of its primary functionality, the skill transmits file paths, natural language instructions, and code blocks to the Morph Apply API. Users should be aware that source code context is sent to an external service for processing.- [EXTERNAL_DOWNLOADS]: The skill makes outbound network requests to the Morph Apply API to perform its file modification tasks.- [PROMPT_INJECTION]: The skill possesses an indirect prompt injection surface due to its role in processing untrusted content and instructions.
Ingestion points: File paths, natural language instructions, and code edits provided via command-line arguments and target file content.
Boundary markers: The skill uses structural markers like // ... existing code ... which are designed for context placement rather than as robust security boundaries.
Capability inventory: The skill has the capability to read files and execute scripts that can modify the local filesystem.
Sanitization: There is no documented sanitization or validation of the instructions or code fragments before they are transmitted to the external API.

morph-apply