Security audit
Security checks for vulnerabilities and agentic risk
This skill is a disclosed Gemini Deep Research client that sends user research requests to Google's API and saves the returned report locally.
Install only if you are comfortable sending research prompts and any selected file-search context to Google's Gemini API. Do not use it with secrets, regulated data, or confidential business material unless that data is approved for this provider, and protect or delete the generated JSON metadata files when they may contain sensitive content.
Credentials or environment variables flow to a network sink. This is a high-confidence indicator of credential exfiltration.
}
while True:
response = requests.get(
f"{API_BASE}/interactions/{interaction_id}",
headers=headers
)
The skill advertises capabilities that involve reading environment variables, making network requests, and writing files, but it does not declare an explicit tool or permission scope. That creates an authorization and review gap: operators and downstream systems cannot easily constrain or audit what the skill is allowed to access, increasing the chance of over-privileged execution or unintended data handling.
The skill encourages optional file-search against user data and states that it saves full interaction metadata, but it does not warn that sensitive internal content may be sent to an external API or persisted locally in JSON output. In a research skill that performs systematic web search and synthesis, this context makes the issue more dangerous because users may supply proprietary reports, competitive intelligence, or personal data without understanding the exposure and retention risks.
The script transmits the user query and optional file-search context to a remote Google API without an explicit privacy notice or confirmation step. Because this skill is specifically designed for deep research and multi-source synthesis, users may provide sensitive internal data, making undisclosed external transmission a meaningful privacy and data-governance risk.
Data is being sent to an external URL. This could be legitimate telemetry or data exfiltration. Manual review is recommended.
"file_search_store_names": [file_search_store]
}]
response = requests.post(
f"{API_BASE}/interactions",
headers=headers,
json=payload
The script automatically writes the full research report and raw JSON response to local disk, which can persist sensitive prompts, proprietary research, file-search-derived content, or model outputs longer than the user expects. In a research skill, users may submit confidential business or technical material, so silent persistence increases data exposure risk on shared machines, repos, backups, or endpoint monitoring systems.
No suspicious patterns detected.