import-skills-to-skilder
Pass
Audited by Gen Agent Trust Hub on Jul 28, 2026
Risk Level: SAFE
Full Analysis
- [DATA_EXFILTRATION]: The skill reads local files from project directories and the user's home folder (e.g.,
~/.claude/skills/) and transmits the contents toapp.skilder.ai. This is the primary functionality of the tool, intended to centralize skill management. - [DATA_EXFILTRATION]: Safety constraints are explicitly included to prevent sensitive data exposure. The instructions require the agent to check for and exclude files containing API keys, tokens, passwords, or
.envcontent before transmission. - [PROMPT_INJECTION]: The skill represents a surface for indirect prompt injection as it ingests untrusted instruction files from local repositories and project directories.
- Ingestion points:
SKILL.mdand companion files from local and project directories (Step 2). - Boundary markers: No specific delimiters or safety warnings are added to the imported content; instructions are imported verbatim (
Step 4). - Capability inventory: Uses the
managetool suite to create skills and content on the remote platform (Step 4). - Sanitization: The skill performs credential scanning but does not sanitize or validate the natural language instructions within the skills (
Constraints). - [SAFE]: The network communication targets the vendor's own infrastructure, and the file access is restricted to the specific task of migrating agent skills as requested by the user.
Audit Metadata