The Agent Skills Directory

[PROMPT_INJECTION]: No malicious instructions, bypass attempts, or safety filter overrides were detected in the skill definitions. All instructions focus on legitimate database management operations.
[CREDENTIALS_UNSAFE]: The skill correctly instructs the agent to prompt the user for connection URIs and authentication tokens. No hardcoded credentials, API keys, or secrets are present in the code or documentation.
[EXTERNAL_DOWNLOADS]: The skill requires the official pymilvus Python package from standard registries. All external links point to official Milvus documentation (milvus.io) or the official GitHub repository for zilliztech.
[COMMAND_EXECUTION]: Shell access (Bash) is requested as part of the allowed-tools to execute Python scripts that interact with the Milvus database. No suspicious, unauthorized, or hidden command execution patterns were found.
[DATA_EXFILTRATION]: No unauthorized data transmission or exfiltration patterns were identified. The code snippets perform database operations (insert, search, delete) against user-defined endpoints, which is the primary purpose of the skill.

milvus