resume-writer

Fail

Audited by Gen Agent Trust Hub on Jul 13, 2026

Risk Level: HIGHCOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
  • [COMMAND_EXECUTION]: The script scripts/Register-ResumeTipsRefreshTask.ps1 establishes a persistence mechanism on Windows hosts using the Register-ScheduledTask cmdlet to execute local scripts periodically.
  • [COMMAND_EXECUTION]: Multiple components, including README.md and the registration scripts, utilize the -ExecutionPolicy Bypass flag to circumvent PowerShell's default security restrictions and execute unsigned or restricted scripts.
  • [PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection due to the ingestion of untrusted external content.
  • Ingestion points: Processes user-provided resume text, content extracted from PDF/DOCX files, and recruiter insights from third-party social media posts as defined in SKILL.md and tips/README.md.
  • Boundary markers: The skill lacks explicit delimiters or instructions (e.g., XML tags or clear 'ignore' blocks) to isolate external input from the internal reasoning and instruction set.
  • Capability inventory: The skill possesses the capability to modify local files (e.g., insights/principles.md), execute shell scripts, and modify system-level tasks.
  • Sanitization: No validation, escaping, or filtering of ingested content is implemented before it is processed by the agent.
  • [COMMAND_EXECUTION]: The Invoke-ResumeTipsRefresh.ps1 script performs arbitrary file system operations, including creating directories and writing Markdown files based on external CSV data.
Recommendations
  • AI detected serious security threats
Audit Metadata
Risk Level
HIGH
Analyzed
Jul 13, 2026, 01:07 AM
Security Audit — agent-trust-hub — resume-writer