The Agent Skills Directory

[PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection because it ingests and parses untrusted content from external endpoints without sufficient sanitization or boundary markers.
Ingestion points: Remote responses from S3, GCS, and Azure Blob storage listings, and NPM/PyPI registry APIs as described in SKILL.md (§6.2 and §8.6).
Boundary markers: No explicit delimiters or instructions to ignore embedded content are used when processing external data.
Capability inventory: The skill provides the agent with capabilities to execute network probes via curl and Invoke-WebRequest, and data decoding via a provided Python script.
Sanitization: No sanitization or validation of the remote response content is performed before parsing.
[COMMAND_EXECUTION]: The skill contains several functional code snippets in Bash, PowerShell, and Python intended for the agent to execute for bucket candidate generation, network probing, and AWS identifier decoding.
[EXTERNAL_DOWNLOADS]: The skill instructs the agent to make network requests to external domains including s3.amazonaws.com, storage.googleapis.com, blob.core.windows.net, registry.npmjs.org, and pypi.org to retrieve metadata and file listings.

cloud-saas-exposure