Security audit
Security checks for vulnerabilities and agentic risk
This skill is a disclosed anime lookup and streaming-link CLI, but users should treat its unofficial streaming links as potentially risky.
Install only if you are comfortable sending anime searches and IDs to AniList and Anikoto and receiving links to unofficial third-party streaming sites. Prefer official/legal streaming providers when available, and avoid opening unfamiliar streaming domains in sensitive browser profiles.
The skill explicitly advertises direct third-party streaming links from domains like megaplay.buzz and anixo.buzz without warning users about legal status, tracking, malware, or other safety risks. This is dangerous because users may be steered to unvetted external services under the implicit trust of the skill, exposing them to privacy harm, malicious content, or copyright-related risk.
The user-facing documentation and trigger wording are written in Indonesian, including activation cues such as 'bertanya', 'minta rekomendasi', and 'cari anime untuk ditonton'. There is no indication that language is optional or that the skill supports user choice, which may violate language/locale policy expectations.
The trigger section uses broad, everyday terms like 'anime', 'episode', 'streaming', and generic recommendation phrases, which can cause the skill to activate for loosely related user requests. Over-broad invocation increases the chance of unintended tool use and unsolicited delivery of third-party streaming links, especially in multi-skill environments where precise routing matters.
The module docstring says the CLI is compatible with Telegram formatting and specifically mentions a --json mode without ANSI codes, which implies machine-readable JSON output when that flag is used. However, in cmd_detail, the --json argument is accepted but ignored: the function always prints the Markdown-formatted fmt_detail output and optional URLs rather than JSON.
The tool sends user-provided search terms and anime identifiers to third-party services (anilist.co, anikotoapi.site, and linked streaming domains) without any disclosure, consent, or privacy notice. In an agent/skill context, this can leak user interests and usage metadata to external operators, especially because one service appears unofficial and the skill also promotes direct streaming links to third-party sites.
The output strings in this section are hard-coded in Indonesian (for example, "Stream di Telegram" and "Tap link di atas"), which imposes a specific language on users without offering a choice. The policy requires avoiding forced language or locale constraints unless the skill documents user choice or a justified regional limitation.
The CLI parser defines --json for the detail subcommand, signaling to users that this mode should return JSON as it does for other commands. But cmd_detail never checks args.json for output format selection and always prints formatted text, creating a direct contradiction between the command interface documentation and implementation.
