forge-architecture

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 Node.js script using a relative path: node ../../../.fullstack-forge/runtime/cli/src/composition-entry.js. It also suggests running CLI commands like forge doctor and forge update all, which implies the presence and execution of unverified local tooling.
  • [PROMPT_INJECTION]: The skill uses imperative language to override the agent's current context, directing it to read and follow a "canonical playbook" located at a relative path (../../../.fullstack-forge/skills/forge-architecture/SKILL.md) rather than the current file.
  • [INDIRECT_PROMPT_INJECTION]:
  • Ingestion points: The agent is instructed to read content from ../../../.fullstack-forge/skills/forge-architecture/SKILL.md and .forge/composition.json.
  • Boundary markers: Absent; there are no instructions to delimit or sanitize the content retrieved from these external files.
  • Capability inventory: The skill has the capability to execute shell commands via Node.js and access the file system via relative paths.
  • Sanitization: None detected for the external instructions or the JSON configuration data being processed.
Audit Metadata
Risk Level
MEDIUM
Analyzed
Aug 6, 2026, 06:48 PM
Security Audit — agent-trust-hub — forge-architecture