The Agent Skills Directory

[COMMAND_EXECUTION]: The skill uses shell commands like find and grep to analyze project structure and identifies training or inference entry points. It executes model tests and validation scripts using python and pytest to confirm successful migration.
[EXTERNAL_DOWNLOADS]: The skill instructs the agent to clone external repositories via git clone and install necessary dependencies using pip install from remote sources as part of the environment setup and testing phases.
[PROMPT_INJECTION]: The skill has an attack surface for indirect prompt injection because it processes untrusted data from external repositories, such as README files and source code, which could contain instructions meant to deceive the agent.
Ingestion points: The skill reads README.md, requirements.txt, setup.py, and various Python source files from user-provided repositories.
Boundary markers: Not explicitly defined; the skill lacks specific delimiters or warnings to ignore instructions found within the project files being analyzed.
Capability inventory: File system searching (find), pattern matching (grep), dependency installation (pip), and arbitrary script execution (python).
Sanitization: No sanitization or validation of the content of external project files is mentioned in the migration instructions.

external-gitcode-ascend-npu-model-migration