The Agent Skills Directory

[DATA_EXFILTRATION]: The skill performs network operations to an external domain api.opencclaw.com. This domain is not recognized as a trusted organization or well-known service in the current security context.
[COMMAND_EXECUTION]: The skill provides examples of using subprocess.run to call the dcatalog CLI tool. This demonstrates a capability for the agent to execute shell commands on the host system.
[PROMPT_INJECTION]: The skill processes metadata and search queries from potentially untrusted sources, which constitutes an indirect prompt injection attack surface.
Ingestion points: Metadata registration (dcatalog register --metadata) and asset search (dcatalog search --query).
Boundary markers: No explicit delimiters or instructions to ignore embedded commands are defined in the skill documentation.
Capability inventory: The skill utilizes subprocess.run for CLI interactions and the requests library for network communication with external APIs.
Sanitization: While the documentation suggests validating JSON with json.loads(), there is no evidence of sanitization to prevent malicious instructions within the metadata strings from influencing the agent's logic.

data-catalog