hunt-source-leak

Fail

Audited by Gen Agent Trust Hub on Jun 13, 2026

Risk Level: HIGHEXTERNAL_DOWNLOADSREMOTE_CODE_EXECUTIONCOMMAND_EXECUTIONDATA_EXFILTRATIONPROMPT_INJECTION
Full Analysis
  • [EXTERNAL_DOWNLOADS]: The skill installs several third-party tools from public registries during execution, including Python packages git-dumper and ds_store, as well as Node.js packages source-map-explorer and unwebpack-sourcemap. These dependencies are not pinned to specific versions.
  • [REMOTE_CODE_EXECUTION]: The skill downloads content from target URLs and pipes it directly into Python interpreters (e.g., using python3 -c for source map extraction and Swagger parsing). While the script logic is defined within the skill, processing unvalidated network data through a local interpreter is a high-risk pattern.
  • [COMMAND_EXECUTION]: Extensive use of shell commands and complex pipelines involving curl, grep, sed, xxd, and tail are used to automate the discovery and extraction of information from target servers.
  • [DATA_EXFILTRATION]: The skill explicitly targets sensitive configuration and source control files (e.g., .env, .env.production, and .git/config) that are likely to contain credentials and internal system information.
  • [PROMPT_INJECTION]: The skill exhibits an indirect prompt injection surface by processing untrusted data from target websites. (1) Ingestion points: Multiple curl calls in SKILL.md; (2) Boundary markers: Absent; (3) Capability inventory: Python script execution and file-write operations in SKILL.md; (4) Sanitization: Absent.
Recommendations
  • HIGH: Downloads and executes remote code from: https://$TARGET/asset-manifest.json - DO NOT USE without thorough review
Audit Metadata
Risk Level
HIGH
Analyzed
Jun 13, 2026, 04:05 PM
Security Audit — agent-trust-hub — hunt-source-leak