The Agent Skills Directory

[SAFE]: The skill consists entirely of Markdown-based domain knowledge and reference documentation. No executable scripts (.py, .js, .sh) or binary files are present within the skill structure.
[SAFE]: Instructional language in SKILL.md and workflow-standalone.md provides legitimate guidance for the AI agent to navigate the reference files. No prompt injection attempts, safety filter bypasses, or system prompt extraction patterns were identified.
[DATA_EXPOSURE]: No hardcoded credentials, API keys, or sensitive environment variable patterns were found. The skill does not access sensitive system paths or user data.
[DATA_EXFILTRATION]: No network operations (curl, wget, fetch) or external communications were detected. All references point to local files within the skill's own directory structure.
[COMMAND_EXECUTION]: No shell commands or subprocess calls are executed by the skill. Code snippets provided in the documentation are HLSL/GLSL pseudo-code intended for graphics programming reference, not for execution on the host system.
[REMOTE_CODE_EXECUTION]: No remote script downloads or piped command execution patterns were found. The skill does not depend on any external third-party package registries.
[OBFUSCATION]: Content is stored in cleartext Markdown. No Base64 encoded payloads, zero-width characters, or homoglyphs were detected. Mathematical formulas and shader property names are documented clearly using standard notations.
[INDIRECT_PROMPT_INJECTION]: While the skill ingests user requirements to provide shader advice, it lacks any dangerous capabilities (file writes, network calls, or code execution) that could be exploited via indirect injection. The attack surface is effectively restricted to informational output.

animation-shader