cloudbase-mcp

Pass

Audited by Gen Agent Trust Hub on Nov 13, 2026

Risk Level: SAFEEXTERNAL_DOWNLOADSCOMMAND_EXECUTIONDATA_EXFILTRATIONINDIRECT_PROMPT_INJECTION
Full Analysis
  • [EXTERNAL_DOWNLOADS]: The skill uses npx to fetch and execute official packages from the npm registry: @cloudbase/npm-launcher@1.0.0 and @cloudbase/cloudbase-mcp@latest. These are legitimate vendor resources provided by Tencent CloudBase for launching the MCP server.
  • [COMMAND_EXECUTION]: The skill provides templates for shell commands using mcporter and npx to list, describe, and call MCP tools. These commands are intended for local environment management and interaction with the CloudBase infrastructure.
  • [DATA_EXFILTRATION]: The file references/issue-report.md contains rigorous instructions for sanitizing debug reports. It mandates the removal of secrets (API keys, SecretIds, SecretKeys, tokens), personal identifiers (phone numbers, emails), and business logic (table names, product names) before any information is shared. It explicitly specifies that envId is kept for troubleshooting but notes it grants no access without credentials.
  • [INDIRECT_PROMPT_INJECTION]: The skill instructs the agent on how to process conversation history and tool outputs to generate bug reports. While this involves processing untrusted data from session context, the skill implements a 'Redaction' framework with four distinct tiers to prevent the leakage of sensitive data or the propagation of malicious intent during the report generation process.
  • [SAFE]: All referenced remote endpoints (tcb-api.cloud.tencent.com and tcb-api.tencentcloud.com) and GitHub repositories are official Tencent CloudBase assets. The instructions emphasize secure credential management, such as using ~/.config/.cloudbase/auth.json with mode 600 instead of storing secrets in configuration files.
Audit Metadata
Risk Level
SAFE
Analyzed
Nov 13, 2026, 04:53 AM