Skills
Navigate
skills/opensearch-project/opensearch-agent-skills/log-analytics/Gen Agent Trust Hub

log-analytics

Fail

Audited by Gen Agent Trust Hub on May 20, 2026

Risk Level: HIGHCREDENTIALS_UNSAFEEXTERNAL_DOWNLOADSCOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
  • [CREDENTIALS_UNSAFE]: The file log-analytics.md contains a hardcoded password value My_password_123!@# in the environment variable configuration table.
  • [EXTERNAL_DOWNLOADS]: The skill utilizes uvx to download and execute opensearch-mcp-server-py (at the latest unversioned state) and duckduckgo-mcp-server from external repositories.
  • [COMMAND_EXECUTION]: The workflow involves executing curl commands with user-provided or environment-sourced credentials and running uvx for MCP server tools.
  • [PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection as it ingests untrusted log content from OpenSearch indices and processes it without sanitization or boundary markers.
  • Ingestion points: Untrusted log documents are retrieved using SearchIndexTool and GenericOpenSearchApiTool as described in log-analytics.md.
  • Boundary markers: There are no specified delimiters or 'ignore embedded instructions' warnings for the log messages processed by the agent.
  • Capability inventory: The agent can execute administrative and query-level API calls via the GenericOpenSearchApiTool.
  • Sanitization: No validation or sanitization of ingested log data is performed before it is analyzed.
Recommendations
  • AI detected serious security threats
Audit Metadata
Risk Level
HIGH
Analyzed
May 20, 2026, 02:13 PM
Security Audit — agent-trust-hub — log-analytics