neodata-financial-search

Pass

Audited by Gen Agent Trust Hub on May 5, 2026

Risk Level: SAFEEXTERNAL_DOWNLOADSCOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
  • [EXTERNAL_DOWNLOADS]: The skill connects to the NeoData API at copilot.tencent.com to retrieve real-time financial data, market quotes, and news articles. This is a well-known service aligned with the skill's core financial research purpose.
  • [COMMAND_EXECUTION]: The skill executes local Python and Shell scripts (scripts/query.py, scripts/query.sh) to interface with the API and process data. These scripts use standard argument parsing to handle user queries.
  • [PROMPT_INJECTION]: The skill retrieves natural language content from external financial articles, which presents a surface for indirect prompt injection.
  • Ingestion points: External financial data and article content are ingested through the API responses handled in scripts/query.py and scripts/query.sh (specifically the docData field).
  • Boundary markers: Absent. The skill instructions do not provide delimiters or specific "ignore instructions" warnings to the agent to treat the retrieved article content as untrusted data.
  • Capability inventory: The skill utilizes Bash and Read tools across its scripts, which provides an execution surface if the agent were to follow malicious instructions embedded within the ingested financial content.
  • Sanitization: Absent. There is no evidence of validation or sanitization of the retrieved text from the external NeoData service before it is integrated into the agent's context.
Audit Metadata
Risk Level
SAFE
Analyzed
May 5, 2026, 12:53 AM
Security Audit — agent-trust-hub — neodata-financial-search