The Agent Skills Directory

[DATA_EXFILTRATION] (LOW): The scripts search_stories.py and fetch_story.py perform network requests to non-whitelisted domains (microsoft.com). Additionally, fetch_story.py accepts arbitrary URLs, which can be used to probe internal or external resources (SSRF surface).
[PROMPT_INJECTION] (LOW): The skill is susceptible to indirect prompt injection because it ingests and processes content from external web pages.
Ingestion points: scripts/fetch_story.py (Line 67) fetches arbitrary content based on user or search-provided input.
Boundary markers: No specific delimiters or safety warnings are used to encapsulate the fetched content before passing it back to the agent.
Capability inventory: The skill is limited to network requests and text parsing; it does not possess dangerous capabilities like direct file-system writing or system command execution.
Sanitization: The skill uses regular expressions to strip HTML tags from the fetched content, providing basic text extraction but no semantic sanitization.
[EXTERNAL_DOWNLOADS] (LOW): The skill requires the requests Python library. While a standard and reputable package, it must be installed from an external repository (PyPI).

ms-customer-stories