research-equity-workflow

Pass

Audited by Gen Agent Trust Hub on Jun 18, 2026

Risk Level: SAFE
Full Analysis
  • [PROMPT_INJECTION]: The skill does not contain instructions designed to override agent behavior, bypass safety guardrails, or reveal system prompts. It processes untrusted user input (ticker symbols) and external tool data, creating a potential surface for indirect injection; however, the skill possesses no high-risk capabilities—such as shell execution, file system writes, or unauthorized network calls—that could be exploited via these inputs.
  • Ingestion points: User-supplied ticker symbols and focus areas (SKILL.md Step 1); analyst consensus, fundamental, and pricing data from external tools (Steps 2-6).
  • Boundary markers: No specific delimiters or instructions to ignore embedded commands are present in the synthesis steps.
  • Capability inventory: The skill is limited to report synthesis and presentation using the qa_ and tscc_ tool suites. No subprocess, exec/eval, or file-writing capabilities are present.
  • Sanitization: No explicit sanitization of input ticker formats or tool output is defined.
  • [DATA_EXFILTRATION]: No access to sensitive local file paths (e.g., SSH keys, AWS credentials) or unauthorized network requests to external domains were identified. Interactions are limited to the described LSEG/Refinitiv tool connectors.
  • [REMOTE_CODE_EXECUTION]: The skill does not perform remote script downloads, package installations, or dynamic code execution via eval or exec functions.
  • [COMMAND_EXECUTION]: No shell command execution or subprocess spawning patterns were found in the workflow definitions.
Audit Metadata
Risk Level
SAFE
Analyzed
Jun 18, 2026, 07:25 PM
Security Audit — agent-trust-hub — research-equity-workflow