The Agent Skills Directory

[PROMPT_INJECTION]: The skill provides structured guidance for code generation and refactoring. No instructions were found that attempt to override system prompts, bypass safety filters, or extract underlying system instructions. The language used is purely instructional and focused on technical best practices.
[DATA_EXPOSURE_AND_EXFILTRATION]: No sensitive file paths, hardcoded credentials, or unauthorized network operations were detected. The code examples use mock functions (e.g., fetchUser, analytics.track) common in development documentation and do not perform any actual data transmission to external domains. All provided URLs are plain-text references to official library documentation.
[OBFUSCATION]: A scan for Base64 encoding, zero-width characters, homoglyphs, and hidden Unicode tags returned no findings. The content is entirely human-readable and uses standard markdown and code block formatting.
[UNVERIFIABLE_DEPENDENCIES_AND_REMOTE_CODE_EXECUTION]: The skill references standard, well-known software libraries including react-hook-form, zod, lodash, and @mui/material. There are no commands for downloading or executing remote scripts (e.g., curl | bash) and no use of dynamic code execution functions like eval() or exec().
[PRIVILEGE_ESCALATION]: The instructions do not involve any system-level commands, administrative privilege requests, or attempts to modify environment permissions.
[PERSISTENCE_MECHANISMS]: No patterns related to shell profile modification, startup task creation, or service installation were found. The skill does not attempt to maintain any state across agent sessions beyond the static documentation provided.
[METADATA_POISONING]: The metadata in SKILL.md and metadata.json accurately reflects the skill's purpose. There are no hidden instructions in tags, author fields, or versioning information.
[INDIRECT_PROMPT_INJECTION]: While the skill defines patterns for processing form data, it does not ingest untrusted external data at runtime. The instructions include best practices for validation and type safety, which generally improve the robustness of the resulting code. The skill itself has no operational capabilities to write files or make network requests.
[TIME_DELAYED_OR_CONDITIONAL_ATTACKS]: No logic was found that gates behavior based on date, time, or environment-specific conditions.
[DYNAMIC_EXECUTION]: The skill does not perform runtime compilation, library injection, or unsafe deserialization. All code snippets are provided as static templates for developer reference.

react-hook-form