The Agent Skills Directory

[CREDENTIALS_UNSAFE]: The skill instructs the agent to launch a search UI by passing the OpenSearch password as a command-line argument (--password <password>) in SKILL.md. This practice exposes the credential to any process or user capable of viewing the system's process list.
[COMMAND_EXECUTION]: The skill invokes local Python scripts using uv run python scripts/opensearch_ops.py to handle deployment tasks and launch the Search UI.
[EXTERNAL_DOWNLOADS]: The skill configures the agent to install and run the opensearch-mcp-server-py package via uvx at runtime. This package is managed by the opensearch-project vendor.
[REMOTE_CODE_EXECUTION]: The agent is instructed to connect to a remote HTTP MCP server at https://mcp.aiven.live/mcp. This enables the remote server to provide tool definitions and results to the agent.
[DATA_EXFILTRATION]: The skill processes cluster logs which represent an untrusted data ingestion surface, potentially allowing for indirect prompt injection or data exposure.
Ingestion points: aiven_project_get_service_logs tool output mentioned in SKILL.md.
Boundary markers: Absent; instructions do not specify wrapping log content in delimiters.
Capability inventory: The skill has access to uv run for subprocess execution and various Aiven management tools.
Sanitization: No sanitization or filtering of log content is described.

aiven-setup