The Agent Skills Directory

[COMMAND_EXECUTION]: The skill performs remote command execution on Alibaba Cloud ECS instances using the official Cloud Assistant run-command API. This is the core functionality of the diagnostic tool and is gated by a check-write-operation.sh hook that requires user confirmation for execution-level operations.
Evidence: mtr_ecs.py generates shell scripts at runtime and transmits them via the ecs:RunCommand API action.
[EXTERNAL_DOWNLOADS]: The skill references and fetches data from external sources for configuration and diagnostics.
Evidence: mtr_common.py provides download instructions for the aliyun-cli from aliyuncli.alicdn.com.
Evidence: mtr_analyze.py makes HTTPS requests to the well-known service ipinfo.io to retrieve ISP and ASN metadata for network hops.
[DATA_EXFILTRATION]: The skill accesses local configuration files to manage cloud credentials.
Evidence: mtr_common.py reads the aliyun CLI configuration file located at ~/.aliyun/config.json to verify the availability of authentication profiles.
[PROMPT_INJECTION]: The skill exhibits an indirect prompt injection surface area by ingesting and parsing untrusted data from external tool outputs.
Ingestion points: mtr_common.py parses free-text user input, and mtr_analyze.py parses text output from MTR and Ping commands.
Boundary markers: The skill does not use high-level delimiters for prompt interpolation but relies on strict regex-based extraction to filter inputs.
Capability inventory: The skill possesses powerful capabilities including remote shell execution (RunCommand) and network access.
Sanitization: Shell commands are constructed using shlex.quote to prevent command injection, and data extraction is limited to specific patterns (IPs, Regions, and MTR metrics).

alibabacloud-mtr-network-diagnosis-customer