x-threads-engagement
Pass
Audited by Gen Agent Trust Hub on Aug 19, 2026
Risk Level: SAFECOMMAND_EXECUTION
Full Analysis
- [COMMAND_EXECUTION]: The workflow involves executing local Node.js utility scripts ("thread-writer.mjs", "audit-threads.mjs") to automate the conversion of JSON plans into markdown threads and perform validation checks. These scripts use standard Node.js modules for file operations and operate locally within the project directory without performing network requests.\n- [INDIRECT_PROMPT_INJECTION]: The skill implements a multi-agent workflow where an "x-auditor" subagent is instructed to evaluate generated files ("thread.md", "engagement.md") derived from user-provided plans. This creates a potential surface for indirect prompt injection, as malicious instructions could be embedded in the thread text. However, the automated script validation and the specific evaluation framework (scorecard) act as significant mitigating factors.\n
- Ingestion points: The "x-auditor" subagent reads generated markdown files containing content originally sourced from "thread-plan.json".\n
- Boundary markers: Not explicitly defined in the generated markdown, though the subagent brief provides a structured evaluation framework.\n
- Capability inventory: Execution of local Node.js scripts and file operations within the directory.\n
- Sanitization: Intermediary scripts perform regex-based validation of tweet roles and lengths, and check for blocklisted fluff and bait terms.
Audit Metadata