The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill possesses a vulnerability surface for indirect prompt injection because it is designed to ingest and process external, potentially untrusted data sources such as contract documents and spend reports.
Ingestion points: The logic in SKILL.md (e.g., analyze_contract_compliance and contract_portfolio_analytics) processes external DataFrames containing supplier and contract information.
Boundary markers: The instructions do not define clear delimiters or provide instructions to the agent to ignore natural language instructions embedded within the data fields.
Capability inventory: Across all provided scripts, capabilities are restricted to mathematical and logical data manipulation using pandas and numpy. No evidence of shell execution, network exfiltration, or file-system write capabilities was found.
Sanitization: There is no evidence of input validation, escaping, or filtering for external content before it is processed by the agent.

contract-management