The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill's core functionality involves ingesting and processing external data, which creates a vulnerability surface for indirect prompt injection attacks.
Ingestion points: The analysis_pipeline function in SKILL.md uses pd.read_csv(filepath) to load external data from the file system into the agent's memory.
Boundary markers: The instructions and code do not implement boundary markers or specific guidance to ensure the agent ignores natural language instructions that might be embedded within dataset values.
Capability inventory: The skill enables file reading, complex statistical processing, and file writing through visualization export functions (e.g., fig.savefig in SKILL.md).
Sanitization: There is no evidence of data sanitization or validation routines to detect or neutralize adversarial instructions hidden within the input files.

data-analysis