The Agent Skills Directory

[SAFE]: No security issues were detected in the analyzed files. The skill acts as a documentation repository for ClickHouse troubleshooting and does not include any executable scripts or active components.
[DATA_EXPOSURE_AND_EXFILTRATION]: The skill follows security best practices by recommending that secrets be generated at runtime and stored in configuration files with restricted permissions (0600). It explicitly avoids including plaintext credentials in the documentation or query logs.
[REMOTE_CODE_EXECUTION]: While the skill mentions commands for installing software via official repositories (e.g., packages.clickhouse.com), it does not contain any patterns of executing remote code from untrusted or unknown sources.
[EXTERNAL_DOWNLOADS]: All external references, such as GitHub repositories (getcolors/*) and domain names (bigconfig.online), are identified as vendor-owned resources belonging to the author or official service providers (AWS, Vultr, Cloudflare). These are documented neutrally and do not represent a threat.
[INDIRECT_PROMPT_INJECTION]: As a documentation-focused skill, it does not possess an active attack surface for indirect prompt injection. It provides static information to assist an agent in troubleshooting technical failures.

clickhouse-replicated