The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill references external resources from trusted domains and well-known services. It suggests using npx to install or run packages from the NPM registry and provides a curl command to download the skill file itself from Netlify's official GitHub repository.
[COMMAND_EXECUTION]: The skill defines a configuration schema that includes a run_script action for test setup and teardown. These actions execute shell commands (e.g., git init, npm install, rm -rf) within the agent's workspace. While this is the intended functionality of the AXIS framework, it represents a powerful capability.
[PROMPT_INJECTION]: The skill is vulnerable to indirect prompt injection (Category 8), as it processes configuration data that could be controlled by an external party.
Ingestion points: The skill reads axis.config.json and various scenario JSON files from the local filesystem.
Boundary markers: None. There are no instructions for the agent to distinguish between its own system instructions and potentially malicious instructions embedded within the JSON files it reads.
Capability inventory: Significant. The skill has the ability to execute arbitrary shell scripts through the run_script field, perform filesystem operations, and interact with network services via MCP.
Sanitization: None. The skill does not describe any validation or escaping of the content read from configuration files before the agent acts upon it.

configure-axis