The Agent Skills Directory

[COMMAND_EXECUTION]: The skill's 'GLOBAL PROTOCOLS' instruct the agent to use an 'ExecutionProxy Interface' with the 'rtk' prefix (e.g., rtk npm test) for terminal actions. While intended for testing and validation ceremony, this provides the agent with the capability to execute shell commands.
[INDIRECT_PROMPT_INJECTION]: The skill processes untrusted data from .canvas files which can contain malicious instructions embedded in node text fields.
Ingestion points: The agent reads and parses existing .canvas files as part of its 'Common Workflows' (SKILL.md).
Boundary markers: The instructions lack explicit boundary markers or warnings to ignore instructions embedded within the JSON content of the canvas nodes.
Capability inventory: The agent has capabilities for file system read/write and terminal command execution (e.g., rtk npm test).
Sanitization: There are no instructions for sanitizing or escaping the text content parsed from the canvas files before it enters the agent's context.

json-canvas