mf-type-check
Pass
Audited by Gen Agent Trust Hub on Mar 15, 2026
Risk Level: SAFECOMMAND_EXECUTIONEXTERNAL_DOWNLOADSPROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill instruction to run
node scripts/type-check.js --context '<MFContext-JSON>'performs direct string interpolation of external JSON data into a shell command. This creates a command injection surface if the JSON context contains shell metacharacters like single quotes. - [COMMAND_EXECUTION]: The skill uses a temporary TS config path read from a local file (
.mf/diagnostics/latest.json) as an argument fornpx tsc --project <tmp-tsconfig>. A malicious project could craft this file to perform argument injection or execute arbitrary commands. - [EXTERNAL_DOWNLOADS]: The troubleshooting workflow involves fetching remote type definitions (
@mf-types.zip) andremoteEntryURLs. Fetching and potentially processing assets from URLs found in untrusted project configurations carries a risk of interacting with malicious external content. - [PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection because it ingests data from local project files (
tsconfig.json,latest.json) and skill context without sanitization or boundary markers, while possessing the capability to modify files and execute shell commands.
Audit Metadata