The Agent Skills Directory

[COMMAND_EXECUTION]: The skill invokes a local script located at <bundle-root>/analysis/run.py using python3 via the Bash tool to perform deep financial analysis. This script is part of the skill bundle and uses a local support library for processing market and account parameters.
[PROMPT_INJECTION]: The skill presents an indirect prompt injection surface through the ingestion of external data provided via the --headline and --headline-file arguments.
Ingestion points: External data enters the agent's context through user-provided news headlines and specific file paths provided as arguments to the execution command.
Boundary markers: The skill's instructions explicitly mandate that any user-supplied headlines or assumptions must be labeled as such and accompanied by a disclaimer that they are not independently verified.
Capability inventory: The skill is granted access to Bash, Read, Glob, and Grep, enabling it to execute local code and read files within the environment.
Sanitization: Security relies on descriptive instructions and the agent's enforcement of provenance rules rather than automated sanitization of the input text.

analysis