The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill ingests untrusted external data (Current Period TB, Prior Period TB, Budget, and Prior Commentary) to perform financial analysis. While the scripts (scripts/compute_variances.py and scripts/anomaly_detect.py) use structured dataframes, the 'Prior Commentary' input could potentially contain instructions aimed at biasing the LLM's classification or explanations.
Ingestion points: SKILL.md defines 'Required inputs' including trial balance files and optional 'prior commentary' which are read and processed by the agent.
Boundary markers: No specific boundary markers or instructions to ignore embedded commands in the input data are defined in the workflow.
Capability inventory: The skill uses pandas for data manipulation, argparse for CLI, and standard file I/O (CSV/Excel). No dangerous capabilities like network access or arbitrary command execution were found.
Sanitization: The skill lacks explicit sanitization or escaping of the user-provided text in the 'Prior Commentary' before it is used to draft new commentary.
[COMMAND_EXECUTION]: The skill utilizes local Python scripts for data processing. These scripts use standard libraries (pandas, argparse, pathlib) and do not perform any risky subprocess calls or shell command execution.

flux-analysis