The Agent Skills Directory

[PROMPT_INJECTION]: The instructions focus entirely on film directing theory. No commands or patterns were found that attempt to override agent behavior, bypass safety guardrails, or extract system prompts.
[DATA_EXFILTRATION]: No network operations (curl, wget, fetch) or sensitive file path references (credentials, .env, SSH keys) were detected. There is no mechanism for data exfiltration.
[OBFUSCATION]: The content is provided in clear text. Analysis detected no Base64 encoding, zero-width characters, homoglyphs, or steganographic patterns hiding malicious payloads or URLs.
[REMOTE_CODE_EXECUTION]: The skill does not include or reference external scripts, package managers (npm, pip), or remote resources. It is entirely composed of static markdown and JSON test cases.
[DYNAMIC_CONTEXT_INJECTION]: The skill does not use the shell-execution syntax (!command) to perform actions at load time.
[INDIRECT_PROMPT_INJECTION]: While the skill processes user input related to film scenes, it lacks any high-risk capabilities (file writing, network access, or command execution) that could be exploited via malicious data injection. The capability surface is limited to text generation.

shot-design-pov-spectrum