workspace-surface-audit

Warn

Audited by Gen Agent Trust Hub on Apr 14, 2026

Risk Level: MEDIUMCREDENTIALS_UNSAFEDATA_EXFILTRATIONPROMPT_INJECTION
Full Analysis
  • [CREDENTIALS_UNSAFE]: The skill is explicitly designed to read and parse sensitive files including .env, .mcp.json, and .claude/settings*.json. These locations are standard for storing API keys, authentication tokens, and private infrastructure configuration.
  • [DATA_EXFILTRATION]: The skill instructs the agent to inventory and report on the existence of specific secret keys (e.g., STRIPE_API_KEY, TWILIO_AUTH_TOKEN). Mapping the presence of these credentials across an environment is a form of sensitive data exposure, even if the values themselves are not displayed.
  • [PROMPT_INJECTION]: The skill ingests untrusted data from the repository, such as README.md and project manifests. This creates an attack surface for indirect prompt injection where malicious files in a repository could attempt to influence the agent's behavior during the audit process.
  • Ingestion points: Reads package.json, README.md, .mcp.json, and other local repository files.
  • Boundary markers: None provided in the skill instructions to separate repository data from agent instructions.
  • Capability inventory: The skill has read access to the entire workspace and environment surface.
  • Sanitization: No explicit sanitization or filtering of the content read from repository files is specified beyond the instruction to hide secret values.
Audit Metadata
Risk Level
MEDIUM
Analyzed
Apr 14, 2026, 12:20 AM
Security Audit — agent-trust-hub — workspace-surface-audit