tear-sheet
Pass
Audited by Gen Agent Trust Hub on Sep 1, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONDYNAMIC_EXECUTIONCOMMAND_EXECUTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill ingests and processes potentially untrusted external data, such as earnings transcripts and company summaries, retrieved through MCP tools. This creates an attack surface where malicious instructions embedded in the external data could influence the agent's behavior during synthesis.\n
- Ingestion points: Data is fetched from S&P Global MCP tools (e.g., earnings transcripts, company summaries, relationship data).\n
- Boundary markers: The skill does not implement delimiters or specific instructions to the model to ignore embedded commands in the ingested data.\n
- Capability inventory: The skill allows for directory creation (
mkdir), file system writes (/tmp/), and Node.js script execution for document generation.\n - Sanitization: There are no provisions for sanitizing or escaping the content retrieved from external sources before it is processed.\n- [DYNAMIC_EXECUTION]: The skill provides Node.js functions and logic to be used in an agent-generated script for document creation. This involves script generation and execution based on provided templates.\n- [COMMAND_EXECUTION]: The instructions direct the agent to use shell commands (
mkdir -p /tmp/tear-sheet/) for managing temporary directories on the local file system.
Audit Metadata