The Agent Skills Directory

[PROMPT_INJECTION]: No malicious override or bypass instructions were detected. Technical terms like 'override' are used strictly in the context of Svelte 5's reactivity features (e.g., overriding derived values).
[DATA_EXFILTRATION]: No sensitive file paths or hardcoded credentials were found. Network operations mentioned in examples (e.g., fetch('/api/search')) are standard development patterns for documentation purposes.
[OBFUSCATION]: No hidden content, zero-width characters, homoglyphs, or encoded commands were detected. The content is primarily technical documentation in Chinese and English.
[REMOTE_CODE_EXECUTION]: No remote scripts or unverifiable dependencies are downloaded or executed. Dependencies mentioned (like svelte/reactivity) are standard parts of the Svelte ecosystem.
[INDIRECT_PROMPT_INJECTION]: While the skill processes user code for refactoring (ingestion point), it serves as a static reference library and does not introduce new autonomous capabilities that would facilitate injection attacks beyond the agent's base environment. No malicious instructions were found embedded in the reference data.

svelte-runes