korean-spell-check
Pass
Audited by Gen Agent Trust Hub on Apr 13, 2026
Risk Level: SAFE
Full Analysis
- [DATA_EXFILTRATION]: The skill transmits user-provided text or local file contents to the official Nara/PNU Korean spell-checker domain (
nara-speller.co.kr). This network communication is the primary intended function of the skill, and the target is a well-known academic service. The documentation explicitly warns users against sending sensitive data to external services. - [COMMAND_EXECUTION]: The skill executes a bundled Python script (
scripts/korean_spell_check.py) to handle file reading, network requests, and result parsing. The script implements conservative request handling, including text chunking and rate limiting (throttling), to ensure reliable operation and respect for the remote service's capacity. - [PROMPT_INJECTION]: As the skill processes arbitrary text from user inputs or local files, it possesses an attack surface for indirect prompt injection.
- Ingestion points: Text or file content processed via the
scripts/korean_spell_check.pyscript's--textor--filearguments. - Boundary markers: None explicitly implemented; the system relies on the structured output format (JSON/Text) defined in the script.
- Capability inventory: The agent summarizes corrections and presents proofread results based on the output of the external service.
- Sanitization: The script extracts content from the external service's HTML responses using regular expressions but does not perform additional sanitization of the text for potential instructions.
Audit Metadata