linkedin-character-limits
Pass
Audited by Gen Agent Trust Hub on Sep 29, 2026
Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTIONEXTERNAL_DOWNLOADSDYNAMIC_EXECUTION
Full Analysis
- [COMMAND_EXECUTION]: The skill instructs the agent to execute a terminal command (
printf %s "$draft" | wc -m) to verify character counts for user content.\n- [DYNAMIC_EXECUTION]: The skill recommends a shell execution pattern where user-provided content ($draft) is interpolated into a command line template.\n - While the template uses double quotes as a best practice, runtime assembly of shell commands using untrusted input remains a technical risk for command injection if not handled by a secure execution environment.\n- [INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and process untrusted user-provided LinkedIn drafts, creating an injection surface.\n
- Ingestion points: User-provided drafts for posts, headlines, and articles (SKILL.md).\n
- Boundary markers: None specified for the draft content.\n
- Capability inventory: Includes instructions for shell command execution (printf, wc).\n
- Sanitization: No instructions provided for escaping or sanitizing draft content before it is processed by the agent or shell utilities.\n- [EXTERNAL_DOWNLOADS]: The skill references external tools and services provided by the vendor, liftli-ai.\n
- Provides instructions for adding a Model Context Protocol (MCP) server from mcp.liftli.ai.\n
- Includes links to additional tools and documentation on the liftli.ai domain.
Audit Metadata