pp-vercel-admin
Warn
Audited by Gen Agent Trust Hub on Aug 17, 2026
Risk Level: MEDIUMCREDENTIALS_UNSAFEEXTERNAL_DOWNLOADSREMOTE_CODE_EXECUTIONCOMMAND_EXECUTIONDATA_EXFILTRATION
Full Analysis
- [EXTERNAL_DOWNLOADS]: The skill instructs the agent to install the
vercel-admin-pp-clitool usingnpxorgo installfrom repositories and packages associated with the vendormvanhorn. These are considered vendor-specific resources. - [DATA_EXFILTRATION]: The CLI tool supports a
--deliver webhook:<url>flag, which allows the agent to route command output—potentially including sensitive data or credentials—to arbitrary external URLs. This functionality represents a high-risk data exfiltration channel. - [COMMAND_EXECUTION]: The skill provides access to the
sandboxes run-session-command, which allows the execution of arbitrary shell commands within a Vercel sandbox environment. This grants the agent a high level of autonomous execution capability. - [CREDENTIALS_UNSAFE]: The skill documentation includes commands for retrieving decrypted environment variables (
env get-shared-var), listing authentication tokens (user list-auth-tokens), and obtaining domain authentication codes (registrar get-domain-auth-code). - [INDIRECT_PROMPT_INJECTION]: The skill processes data from external sources, including Vercel API responses and files from sandboxes, which could contain malicious instructions designed to influence the agent's behavior.
- Ingestion points: Vercel API responses, files retrieved via
sandboxes read-session-file, and local feedback logs in~/.local/share/vercel-admin-pp-cli/feedback.jsonl. - Boundary markers: None present in the instructions to delimit or sanitize external data.
- Capability inventory: The skill can execute shell commands in sandboxes, perform network requests to the Vercel API, and write files or send data to webhooks via the delivery mechanism.
- Sanitization: No evidence of input validation or content escaping for processed data.
Audit Metadata