The Agent Skills Directory

[COMMAND_EXECUTION]: The skill's primary logic resides in a Python script (scripts/main.py) that performs high-risk file system operations. Specifically, the move and batch commands utilize shutil.move and shutil.rmtree to relocate or delete directories based on user-supplied paths. This provides the capability to manipulate arbitrary directories, potentially affecting sensitive system or personal data if incorrect paths are targeted.
[PROMPT_INJECTION]: The skill provides an attack surface for indirect prompt injection through the processing of external manifest files in the batch command.
Ingestion points: Path strings and manifest file contents read in scripts/main.py via the batch command arguments.
Boundary markers: None; the manifest file is parsed line-by-line using tab delimiters without any instructions to ignore embedded commands or verify project boundaries.
Capability inventory: Powerful filesystem manipulation capabilities, including moving, merging, and deleting directories.
Sanitization: Absent; while the script resolves paths, it does not validate them against a list of safe or expected project roots, allowing access to any path the current user can modify.

migrate-sessions