The Agent Skills Directory

[COMMAND_EXECUTION]: The skill uses shell commands to perform financial calculations and verify data using a local script. Specifically, it executes python3 tools/financial_rigor.py with arguments derived from report content.
[INDIRECT_PROMPT_INJECTION]: The skill has a vulnerability surface for indirect prompt injection as it processes data from external files that may be untrusted.
Ingestion points: The skill reads financial data and evidence from files located in the reports/ directory (e.g., reports/{company}-thesis.md).
Boundary markers: No explicit delimiters or instructions are provided to the agent to treat the content of the reports as potentially untrusted or to ignore embedded instructions.
Capability inventory: The skill possesses the capability to execute shell commands via python3 subprocesses in Step A3.
Sanitization: There is no evidence of sanitization or escaping of the values (like {price}, {EPS}, or {JSON}) extracted from the reports before they are interpolated into the shell command strings.

thesis-drift