Joe Security Sandbox
|
|
| Slug |
joe-security-sandbox |
| Category |
Sandbox |
| Maturity |
Beta |
| Auth Type |
API key |
MCP integration with Joe Sandbox (cloud or on-prem) using the Joe Sandbox API v2. Lets a Prospector Studio agent submit files, URLs, or full cookbooks for deep malware analysis; check status; retrieve reports; and search prior analyses by hash, filename, or URL.
This plugin is read-write — it can submit and delete samples and analyses. Bind it to agents only when that level of authority is intended.
- Submit files, URLs, file URLs, and custom cookbooks for analysis.
- List and search prior submissions and analyses; retrieve full reports.
- Delete submissions and analyses.
- Check sandbox health, server info, supported analysis systems, languages, locales, and LIA countries.
- Read account info including remaining quota and permissions.
| Header |
Required |
Description |
X-JoeSandbox-Base-Url |
yes |
Joe Sandbox instance URL (e.g. https://jbxcloud.joesecurity.org or your on-prem URL). |
X-JoeSandbox-Api-Key |
yes |
Joe Sandbox API key. |
The Joe Sandbox API key is issued per-user from the account profile and inherits the owning account's permissions, quotas, available analyzers, sample-submission types, and report-download formats. There are no separate scopes on the key itself — control what the agent can do by issuing the key under a service account whose tier and configured Web API access flags match the intended behavior. Reference: Joe Sandbox Web API guide.
* — outbound is gated only by the base-URL header. Use a tightly-scoped network policy upstream of Studio if access must be restricted to a specific Joe Sandbox tenant.
| Tool |
Action |
Purpose |
submit_sample |
write |
Submit a file for analysis. |
submit_url |
write |
Submit a URL for analysis. |
submit_sample_url |
write |
Submit a file URL for the sandbox to download and analyze. |
submit_cookbook |
write |
Submit a custom cookbook. |
get_submission_info |
read |
Submission info including all analysis IDs. |
list_submissions |
read |
List all submissions. |
delete_submission |
write |
Delete a submission and its analyses. |
| Tool |
Action |
Purpose |
get_analysis_info |
read |
Analysis info by web ID. |
list_analyses |
read |
List analyses with pagination. |
search_analyses |
read |
Search by hash, filename, or URL. |
download_report |
read |
Download a report or resource. |
delete_analysis |
write |
Delete an analysis. |
| Tool |
Action |
Purpose |
server_online |
read |
Server health check. |
server_info |
read |
Server information and version. |
server_systems |
read |
Available analysis systems and configurations. |
server_lia_countries |
read |
Available localized internet anonymization countries. |
server_languages_and_locales |
read |
Available languages and locales. |
account_info |
read |
Account info including quota and permissions. |
{
"name": "submit_url",
"arguments": {
"url": "https://example.com/suspicious-page",
"analysis_systems": ["w10x64"]
}
}
- Submissions consume quota. Joe Sandbox accounts are quota-limited. Have agents consult
account_info before bulk submissions, and prefer search_analyses to reuse prior reports for known hashes.
- Asynchronous flow. Submission tools return a web ID; results land later. Agents should poll with
get_analysis_info rather than blocking on submission.
- Open allowed-hosts list. The plugin's allowed_hosts is
*; the base-URL header is what scopes outbound. Make sure the header is set per-conversation so an agent can't redirect calls.
- Reference docs. Joe Sandbox API v2.