fittings-doc-url-to-markdown
Fail
Audited by Gen Agent Trust Hub on Aug 17, 2026
Risk Level: HIGHREMOTE_CODE_EXECUTIONCREDENTIALS_UNSAFECOMMAND_EXECUTION
Full Analysis
- [REMOTE_CODE_EXECUTION]: The skill fetches an external script from
https://fittings.sh/skill/call.mjsand executes it locally usingnode. This pattern allows the execution of arbitrary remote code that bypasses version control or package manager auditing. - [CREDENTIALS_UNSAFE]: The instructions direct the user to
export FITTINGS_PRIVATE_KEY=0x..., placing a raw cryptographic private key into the shell environment. This practice exposes sensitive credentials to potential logging, process inspections, and other forms of local exposure. - [COMMAND_EXECUTION]: The skill provides instructions for executing shell commands (
curl,node) to download files and interact with the vendor's API. - [INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest data from arbitrary public URLs and return it as Markdown to the agent. This ingestion point lacks explicit sanitization or boundary markers in the description, making the agent susceptible to instructions embedded in the processed web content.
Recommendations
- AI detected serious security threats
Audit Metadata