The Agent Skills Directory

[CREDENTIALS_UNSAFE]: The skill instructs users and the agent to permanently store sensitive Tencent Cloud credentials (TENCENTCLOUD_SECRET_ID, TENCENTCLOUD_SECRET_KEY) by appending export commands to shell configuration files such as ~/.zshrc, ~/.bashrc, or Windows environment variables. Storing credentials in plaintext within shell profiles is a persistence risk.
[COMMAND_EXECUTION]: The check_env.py script uses the Python subprocess module to execute an external CLI tool named clawhub with the inspect command to verify skill versions. The skill also relies on tccli for performing administrative cloud operations.
[PROMPT_INJECTION]: The skill contains persona-enforcement instructions in SKILL.md (the 'Identity Statement' section) that direct the agent to only use the name 'AndonQ' and disregard other labels. Furthermore, the skill processes untrusted data from ticket comments and customer service responses; when combined with its tccli capabilities, this creates a surface for indirect prompt injection where instructions embedded in tickets could influence agent actions.
[DATA_EXFILTRATION]: The skill retrieves environment variables containing high-privilege credentials and transmits them via signed requests (TC3-HMAC-SHA256) to various Tencent Cloud API endpoints, including tandon.tencentcloudapi.com and andon.cloud.tencent.com. While these are official vendor domains, the handling of such sensitive data is a critical security function.

andonq