tencent-cloud-cos
Fail
Audited by Gen Agent Trust Hub on Aug 16, 2026
Risk Level: HIGHCOMMAND_EXECUTIONCREDENTIALS_UNSAFEEXTERNAL_DOWNLOADSPROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The scripts/setup.sh script uses the eval command with unquoted and unsanitized variables (SECRET_ID, SECRET_KEY, BUCKET, REGION) derived from user input, allowing for arbitrary command execution.
- [CREDENTIALS_UNSAFE]: Tencent Cloud credentials (SecretId, SecretKey) are stored in plain text in ~/.mcporter/mcporter.json and ~/.cos.conf. Although file permissions are restricted to 600, storing secrets in plain text is a significant risk.
- [EXTERNAL_DOWNLOADS]: The skill downloads and installs Node.js packages like cos-mcp and mcporter from external registries. The install.sh script uses global installation flags which increases the attack surface.
- [PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection as it ingests untrusted data from Tencent COS. Ingestion points: list_files and search_by_text in scripts/cos_wrapper.py. Boundary markers: None. Capability inventory: File system access, network operations, and shell command execution. Sanitization: None.
- [PRIVILEGE_ESCALATION]: The install.sh script promotes global installation of packages, which often requires and encourages the use of elevated privileges like sudo.
Recommendations
- AI detected serious security threats
Audit Metadata