sequential-thinking
Pass
Audited by Gen Agent Trust Hub on Oct 6, 2026
Risk Level: SAFECOMMAND_EXECUTIONPERSISTENCEINDIRECT_PROMPT_INJECTION
Full Analysis
- [SAFE]: The skill provides a systematic reasoning framework for breaking down complex tasks into manageable, reflective steps.
- [SAFE]: All external resource references point to official, trusted repositories belonging to the Model Context Protocol (MCP) project.
- [COMMAND_EXECUTION]: Local command execution is restricted to the skill's own Node.js helper scripts used for formatting and validation of the reasoning process.
- [PERSISTENCE]: The skill maintains a local session history file to track reasoning sequences, which is standard behavior for stateful problem-solving tools.
- [INDIRECT_PROMPT_INJECTION]: While the skill processes reasoning thoughts, the potential attack surface is minimal: (1) Ingestion occurs through internal processing scripts; (2) Thoughts are delimited by clear 'Thought X/Y' markers; (3) Capabilities are limited to local history logging; (4) Input validation ensures thoughts are processed as strings.
Audit Metadata