sequential-thinking
Pass
Audited by Gen Agent Trust Hub on Oct 4, 2026
Risk Level: SAFECOMMAND_EXECUTIONEXTERNAL_DOWNLOADS
Full Analysis
- [COMMAND_EXECUTION]: The entry script
scripts/main.cjsuseschild_process.spawnto run a local Python executor script (executor.py). The execution is handled securely withshell: falseandwindowsHide: true, preventing common command injection vulnerabilities associated with shell execution. - [EXTERNAL_DOWNLOADS]: The skill utilizes official resources from the Model Context Protocol ecosystem, specifically the
mcpPython package and the@modelcontextprotocol/server-sequential-thinkingNode package vianpx. These are well-known and standard tools for structured AI reasoning. - [INDIRECT_PROMPT_INJECTION]: The skill processes structured 'thought' steps to manage the model's internal reasoning. While it has access to tools like
BashandRead, the input schema is strictly typed, and the primary data flow is internal to the agent's chain of thought, presenting a low surface for indirect injection.
Audit Metadata