The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill ingests data from external observability sources through tools like search-logs, search-traces, and get-distributed-trace (defined in kubesense-logs/SKILL.md and kubesense-traces/SKILL.md). This telemetry data, particularly log bodies and span attributes, could contain malicious instructions. The skill provides write capabilities, such as create-alert and create-dashboard (kubesense-alerts/SKILL.md, kubesense-dashboards/SKILL.md), which could be targeted by such injections. No specific sanitization or boundary marking for ingested telemetry content is explicitly detailed in the instructions.
[EXTERNAL_DOWNLOADS]: The README.md and SKILL.md files provide instructions to install the skill set using npx skills add kubesense-ai/kubesense-mcp-skills. This targets the vendor's (kubesense-ai) own GitHub repository. Per established rules, this is considered a safe reference to the author's own infrastructure.
[COMMAND_EXECUTION]: The skill includes tools for executing raw database queries, specifically execute-sql (ClickHouse SQL) and execute-spl (Log Piped Query Language) in kubesense-sql/SKILL.md and kubesense-spl/SKILL.md. While these queries are executed on the KubeSense server, they provide the agent with powerful data manipulation and retrieval capabilities within the scope of the observability platform.

kubesense-skills