The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill ingests untrusted text data from external financial news and company summaries, creating a potential surface for indirect prompt injection.
Ingestion points: External data is retrieved in exports.py via get_news (news titles and summaries) and get_company_info (business summaries).
Boundary markers: The skill does not implement explicit delimiters or instruction-ignore warnings when returning this content to the agent context.
Capability inventory: The skill possesses network access via the yfinance library and the ability to load local modules.
Sanitization: No sanitization or filtering is performed on the retrieved text to neutralize potential injection patterns before it is processed by the agent.
[DYNAMIC_EXECUTION]: The skill dynamically loads another module from the local file system at runtime.
Evidence: In exports.py, the skill utilizes importlib.util.spec_from_file_location and exec_module to load the twelvedata skill from the hardcoded local path /data/workspace/skills/twelvedata/exports.py.
Context: This is documented as a necessary pattern for cross-skill integration within the platform's workspace to avoid module naming collisions.

us-stock