image-upload
Warn
Audited by Gen Agent Trust Hub on Mar 20, 2026
Risk Level: MEDIUMEXTERNAL_DOWNLOADSDATA_EXFILTRATIONCOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
- [EXTERNAL_DOWNLOADS]: The skill automatically fetches a temporary API key from https://images.labnocturne.com/key if the $LABNOCTURNE_API_KEY environment variable is missing.\n- [DATA_EXFILTRATION]: Local files are read and transmitted to https://images.labnocturne.com/upload. This pattern poses a risk of sensitive data exposure if an attacker tricks the user into providing paths to sensitive files (e.g., keys or config) that have been renamed to appear as supported image extensions.\n- [COMMAND_EXECUTION]: The skill constructs shell commands (curl, ls) using variable substitution from user-supplied paths and data retrieved from remote network requests.\n- [PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection because it processes untrusted data (local file paths) and possesses capabilities to read files and perform network operations.\n
- Ingestion points: User-provided argument in SKILL.md.\n
- Boundary markers: Absent; the skill does not use delimiters or instructions to ignore content within the file path context.\n
- Capability inventory: File existence verification with ls and file upload via curl (SKILL.md).\n
- Sanitization: Implements a whitelist of allowed file extensions (jpg, jpeg, png, gif, webp, svg).
Audit Metadata