The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: Connects to official Alibaba Cloud DashScope API endpoints for model discovery and inference operations. This interaction is essential for the skill's stated purpose and targets well-known service infrastructure.
[COMMAND_EXECUTION]: Provides implementation patterns using standard CLI tools curl and jq to verify endpoint health and process API responses.
[CREDENTIALS_UNSAFE]: Correctly advises users to manage sensitive API keys via environment variables and explicitly warns against storing credentials in local markdown or configuration files.
[DATA_EXFILTRATION]: Manages local state within the ~/qwen/ directory to persist user deployment preferences and routing notes. This data is handled locally and is not transmitted to unauthorized external endpoints.
[PROMPT_INJECTION]: The skill acts as a bridge for processing user prompts and local context through external LLM endpoints.
Ingestion points: User prompt content and local memory files stored in ~/qwen/memory.md.
Boundary markers: No explicit markers are enforced for raw prompts, but the skill provides extensive guidance on using strict schemas for tool-calling.
Capability inventory: Performs network requests to authorized DashScope APIs via curl and reads local deployment metadata.
Sanitization: Recommends the use of jq for validating API response shapes and suggests a two-pass reasoning/normalization pattern to ensure reliable output for automated workflows.

Qwen