✓
Purpose & Capability
Name/description (emoji picker) matches the provided files and behavior: a catalog and guidance for selecting emoji. There are no unrelated binaries, env vars, or config paths requested.
ℹ
Instruction Scope
Runtime instructions simply direct the agent to consult the bundled references/emoji-catalog.md and pick 1–3 emoji; they do not ask the agent to read other system files, call external endpoints, or access credentials. Note: the pre-scan flagged unicode-control-chars in SKILL.md — this can be benign (ZWJ and other invisible characters are used in emoji sequences), but you may want to inspect the raw files for hidden control characters if you are concerned about prompt-injection attempts.
✓
Install Mechanism
No install spec and no code files. Instruction-only skills write nothing to disk and do not fetch external artifacts, which is the lowest-risk install profile.
✓
Credentials
The skill requests no environment variables, credentials, or config paths. That is proportionate for an emoji-picker catalog.
✓
Persistence & Privilege
always:false and no special privileges requested. The skill can be invoked by the agent (default), which is appropriate for an interactive helper of this type.
Scan Findings in Context
[unicode-control-chars] expected: The scanner found invisible/unicode control characters. For an emoji catalog this is often benign (zero-width joiner U+200D and similar are used to compose multi-codepoint emoji). However, invisible characters can also be used for prompt-injection or to obfuscate content—inspect the raw text if you want to be certain.
Assessment
This skill is coherent and low-risk: it only includes documentation (an emoji catalog) and gives simple instructions to consult that catalog. It does not request credentials, install software, or contact external services. The only scanner finding was invisible/unicode control characters—likely harmless here because emoji sequences use zero-width characters, but if you want extra assurance open the SKILL.md and references file in a plain-text editor (showing invisible characters) to verify there are no hidden URLs, commands, or unexpected text. If you plan to allow autonomous agent invocation broadly, remember the agent could call this skill automatically when composing messages, but that is normal and not a security issue for this catalog.
