forge-privacy

Warn

Audited by Gen Agent Trust Hub on Aug 6, 2026

Risk Level: MEDIUMCOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
  • [COMMAND_EXECUTION]: The skill directs the agent to execute the command node ../../../.fullstack-forge/runtime/cli/src/composition-entry.js. This allows the execution of arbitrary logic from a file located outside the skill's own directory.
  • [PROMPT_INJECTION]: The skill exhibits a surface for indirect prompt injection by instructing the agent to dynamically ingest instructions from a 'canonical playbook' and paths specified in a configuration file.
  • Ingestion points: Instructions are read from ../../../.fullstack-forge/skills/forge-privacy/SKILL.md and the .forge/composition.json file.
  • Boundary markers: No explicit delimiters or instructions are provided to the agent to ignore potentially malicious directions within these external files.
  • Capability inventory: The agent is granted the capability to execute shell commands (via node) and read local files.
  • Sanitization: The skill does not implement any validation or sanitization of the content retrieved from external sources before processing it as instruction.
Audit Metadata
Risk Level
MEDIUM
Analyzed
Aug 6, 2026, 06:48 PM
Security Audit — agent-trust-hub — forge-privacy