The Agent Skills Directory

[COMMAND_EXECUTION]: The skill executes the qwen CLI to perform tasks. In delegation mode, it utilizes the --yolo flag, which is explicitly documented in references/cli-qwen.md as a mechanism to auto-approve all actions. This grants the external model significant autonomy to perform system-level changes without per-action user confirmation.
[PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection (Category 8) due to the way it handles untrusted data:
Ingestion points: In references/shared-procedure.md (Step 4), the skill reads project context including design documents, git diff outputs, and implementation files.
Boundary markers: The prompt construction logic in Step 5 does not utilize structural delimiters or markers to isolate the ingested file content from the system instructions, increasing the risk that malicious content in a project file could override agent behavior.
Capability inventory: The skill has the capability to execute commands on the host system via the qwen CLI wrapper, especially when the --yolo autonomy flag is active.
Sanitization: No sanitization or validation routines are specified for the content gathered from local files before it is passed to the external model.

qwen