workspace-surface-audit

Warn

Audited by Gen Agent Trust Hub on Apr 1, 2026

Risk Level: MEDIUMDATA_EXFILTRATIONPROMPT_INJECTION
Full Analysis
  • [DATA_EXFILTRATION]: The skill instructions direct the agent to inspect .env files to identify active services. Reading sensitive files containing credentials and secrets into the agent context constitutes a data exposure risk, even with the explicit instruction to only surface key names and avoid printing values.
  • [PROMPT_INJECTION]: The skill processes untrusted content from various repository files, making it vulnerable to indirect prompt injection.
  • Ingestion points: package.json, README.md, .mcp.json, .lsp.json, and other workspace configuration files (SKILL.md).
  • Boundary markers: Absent; the skill does not use delimiters or instructions to ignore potential commands embedded in the audited files.
  • Capability inventory: Read-only filesystem access for auditing and inventory purposes.
  • Sanitization: Absent; the data from analyzed files is used directly to formulate architectural recommendations.
Audit Metadata
Risk Level
MEDIUM
Analyzed
Apr 1, 2026, 09:47 AM