The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill uses uvx to download and execute the ouroboros-ai[tui] package from the Python Package Index (PyPI) if the ouroboros command is not found in the system PATH.
[COMMAND_EXECUTION]: The skill executes multiple shell commands to manage the system configuration, including ouroboros config, ouroboros config show, and ouroboros config set to modify ~/.ouroboros/config.yaml.
[COMMAND_EXECUTION]: In Branch B, the skill instructs the agent to bind a web server to 0.0.0.0, which exposes the configuration interface to all network interfaces on the host machine. While intended for remote access, this increases the attack surface if the network environment is not trusted.
[INDIRECT_PROMPT_INJECTION]: The agent processes and relays output from the ouroboros CLI tools back to the user, creating a surface where malicious content in configuration files could influence agent behavior. 1. Ingestion points: Output from ouroboros config show and CLI validation error messages. 2. Boundary markers: None explicitly defined for tool output processing. 3. Capability inventory: File system access (reading/writing ~/.ouroboros/config.yaml) and network socket creation. 4. Sanitization: The skill relies on the tool's internal validation, but does not specify sanitization of the output before relaying it to the agent's context.

ouroboros-config