The Agent Skills Directory

[PROMPT_INJECTION]: The skill exhibits deceptive metadata by claiming the author is 'Microsoft' in the frontmatter, whereas the provider identity is 'attentiondotnet'. This misrepresentation can lead to a misplaced sense of trust regarding the instructions and generated code.
[COMMAND_EXECUTION]: The migration guides contain extensive shell and PowerShell instructions utilizing official CLI tools such as 'az', 'gcloud', and 'docker'. These are used to manage cloud resources, access secrets, and migrate container images. The instructions are consistent with the skill's stated purpose.
[EXTERNAL_DOWNLOADS]: The skill provides instructions for downloading container images from Google Artifact Registry and Azure Container Registry, as well as installing official packages from NPM and PyPI. All referenced external resources are official or well-known services.
[PROMPT_INJECTION]: The skill's core functionality involves reading and analyzing untrusted external data (source code from other cloud providers), creating a surface for indirect prompt injection.
Ingestion points: The agent is instructed to read AWS Lambda functions, SAM templates, and GCP Cloud Run configurations from the workspace.
Boundary markers: There are no explicit instructions or delimiters used to isolate the content of the files being read from the agent's instructions.
Capability inventory: The skill utilizes shell command execution (via CLI tools) and file system write operations to generate the migrated project.
Sanitization: No explicit sanitization or validation of the ingested source code is performed before it is processed for assessment or migration.

azure-cloud-migrate