skills/opensearch-project/opensearch-agent-skills/solr-opensearch-migration-advisor/Gen Agent Trust Hub
solr-opensearch-migration-advisor
Warn
Audited by Gen Agent Trust Hub on Apr 27, 2026
Risk Level: MEDIUMCOMMAND_EXECUTIONDATA_EXFILTRATIONREMOTE_CODE_EXECUTIONEXTERNAL_DOWNLOADS
Full Analysis
- [COMMAND_EXECUTION]: The session management logic in
scripts/storage.pyis vulnerable to path traversal. Thesession_idparameter is used directly inos.path.jointo construct file paths (e.g.,sessions/<session_id>.json) for reading and writing data without sanitization. This could be exploited to manipulate files outside the intended directory. - [DATA_EXFILTRATION]: Due to the path traversal vulnerability in the storage backend, there is a risk of unauthorized reading of local JSON files. Additionally, the skill transmits user queries to an external AWS service (
https://knowledge-mcp.global.api.aws) to fetch documentation. While this is a feature of the skill, it represents a data transmission to a remote endpoint. - [REMOTE_CODE_EXECUTION]: The Dockerfile configuration in
setup/docker/claude/performs a remote download and piped execution (curl | bash) of an installation script fromhttps://claude.ai/install.sh. This pattern is an identified security risk, although the source is a trusted service provider. - [EXTERNAL_DOWNLOADS]: The skill references several external resources and APIs, including the AWS Knowledge MCP Server and Anthropic's Claude installation script. All identified external connections are directed at well-known, trusted organizations.
Audit Metadata