forge-frontend

Warn

Audited by Gen Agent Trust Hub on Aug 6, 2026

Risk Level: MEDIUMCOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
  • [COMMAND_EXECUTION]: The skill instructs the agent to execute a local Node.js script using the command node ../../../.fullstack-forge/runtime/cli/src/composition-entry.js frontend compose --root <repository-root> --json. This involves spawning a subprocess to run code that is not contained within the skill itself.
  • [COMMAND_EXECUTION]: The instructions suggest running administrative tools such as forge doctor and forge update all if the installation is perceived as damaged, which grants the agent the ability to execute framework-level maintenance commands.
  • [PROMPT_INJECTION]: The skill redirects the agent to read and follow a "canonical playbook" from an external local path (../../../.fullstack-forge/skills/forge-frontend/SKILL.md) and load further instruction sets defined in a configuration file (.forge/composition.json). This creates a surface for indirect prompt injection where the agent's behavior is determined by data ingested from the local file system without explicit sanitization or boundary markers.
Audit Metadata
Risk Level
MEDIUM
Analyzed
Aug 6, 2026, 06:48 PM
Security Audit — agent-trust-hub — forge-frontend