vibe-manifest-generator

Pass

Audited by Gen Agent Trust Hub on Mar 10, 2026

Risk Level: SAFEPROMPT_INJECTION
Full Analysis
  • [PROMPT_INJECTION]: Indirect prompt injection surface detected. The skill generates core instructions for AI agents by analyzing and extracting data from project files that may be under attacker control.
  • Ingestion points: The Project Analysis phase in SKILL.md extracts information from package.json, README.md, and scripts.
  • Boundary markers: The skill does not generate files with protective delimiters or instructions to ignore embedded commands within the extracted project data.
  • Capability inventory: Manifest templates like AGENTS.md.example promote the use of high-capability tools including the shell subagent and GitHub CLI for PR management.
  • Sanitization: There is no evidence of sanitization or filtering of the extracted project metadata before it is written into the CLAUDE.md and .cursorrules files.
Audit Metadata
Risk Level
SAFE
Analyzed
Mar 10, 2026, 02:06 AM