deep-research
Warn
Audited by Gen Agent Trust Hub on Jul 28, 2026
Risk Level: MEDIUMEXTERNAL_DOWNLOADSREMOTE_CODE_EXECUTIONCOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
- [EXTERNAL_DOWNLOADS]: The installer script fetches components from external GitHub repositories and downloads the
pyyamlPython package from PyPI to support JSON validation and report generation. - [REMOTE_CODE_EXECUTION]: The skill's setup process involves cloning a remote repository and copying scripts into the agent's internal configuration directories, effectively deploying external logic into the agent's execution environment.
- [COMMAND_EXECUTION]: During the report phase, the skill dynamically generates a Python script (
generate_report.py) in the working directory and executes it. This script is built using data collected from external research, which introduces a risk if that data contains malicious content intended to manipulate the script's execution. - [PROMPT_INJECTION]: The workflow involves ingesting large amounts of untrusted data from the web using search tools. This data is merged into prompts and saved to structured files without sanitization or the use of boundary markers, providing an entry point for indirect prompt injection attacks.
Audit Metadata