The Agent Skills Directory

[PROMPT_INJECTION]: The skill instructs the agent to treat files such as AGENTS.md and node_modules/weapp-vite/dist/docs/ as "AI contracts" (SKILL.md, agents/openai.yaml).
Ingestion points: AGENTS.md and node_modules/weapp-vite/dist/docs/ serve as external sources of instructions.
Boundary markers: There are no boundary markers or explicit instructions provided to the agent to disregard potentially malicious instructions embedded within these files.
Capability inventory: The skill grants the agent authority to execute CLI commands (wv, weapp-vite, weapp-ide-cli) that can modify projects or upload code to the IDE.
Sanitization: No sanitization or validation logic is defined for the content ingested from these documentation and contract files.
[COMMAND_EXECUTION]: The skill relies on the execution of various local command-line tools to perform its tasks.
Evidence: Instructions include running wv prepare, wv screenshot, wv compare, and wv ide logs to manage the development and debugging process (SKILL.md, references/ide-command-playbook.md).
Context: These commands are standard for the developer's workflow but could be targeted by indirect injections found in the project's documentation files.

weapp-vite-best-practices