In short
AutoHunter is an automated reconnaissance agent for headless bug-bounty work. It pulls API routes and query parameters out of a site's JavaScript files and queues them for fuzzing by other tools. Odin Research found it on 21 internet-facing servers, most on its default control port 18800, with one confirmed controller.
Key takeaways
- What it is
- A task-specialized recon agent that mines API routes and query parameters, one of seven AI vulnerability agent families in Monograph 20.
- How it works
- It reads JavaScript, extracts endpoints and parameters, and queues them so a fuzzer or scanner can test each one.
- Why it matters
- Every undocumented API route in your front-end code becomes a test target within minutes. Good API inventory is the defense.
- What Odin Research saw
- 21 AutoHunter endpoints on 21 hosts, 14 on port 18800, mostly on Chinese and Hong Kong cloud hosting. One controller is confirmed.
- What to do first
- Inventory and retire unused API routes, and watch for bursts of parameter probing against your web apps.
What is AutoHunter?
AutoHunter is an automated reconnaissance agent built for headless bug-bounty work. It extracts API routes and query parameters from a website's JavaScript files and queues them for fuzzing. It does one step of vulnerability work well: finding places where a test request can land.
Odin Research describes AutoHunter in Monograph 20 as one of seven specialized AI vulnerability agents. Earlier AI offense tools tried to do everything from one console. These newer tools each automate a single slow task.
AutoHunter sits in the surface-discovery stage, next to Cairn. Cairn maps subdomains, certificates and services. AutoHunter goes one level deeper and maps the routes and parameters inside a web application.
We could not find an official public project page or documentation for AutoHunter. What follows comes from what Odin Research observed on servers running it, and from general knowledge of how this kind of reconnaissance works.
How AutoHunter works
Modern web front ends ship a lot of JavaScript. That code often names every API route the application calls, plus the parameters each one accepts. Some of those routes are old, hidden or never meant for public use.
- Collect. The agent fetches a target's pages and JavaScript bundles.
- Extract. It pulls out URL paths, API routes and query parameter names.
- Queue. It hands that list to a recon queue so other tools can fuzz each parameter.
The queue step leaves a network trace. Monograph 20's signature for AutoHunter dispatch looks for an HTTP POST to an /api/autohunter/ route with a target_scope field in the body.
MITRE ATT&CK files this kind of activity under Active Scanning (T1595). Its Wordlist Scanning (T1595.003) sub-technique covers crawling and brute-forcing to find hidden content. AutoHunter's twist is that it builds the list from your own code rather than a generic wordlist.
Why AutoHunter matters to defenders
AutoHunter is dual-use. Bug-bounty hunters and contracted testers run recon tools like it every day. A server running AutoHunter is not proof of hostile intent.
The risk is speed. Monograph 20 models a chain of specialized agents. One family, AutoCVE, turns new advisories into HTTP checks. Discovery agents such as AutoHunter supply the live routes where those checks can land. Container tools such as Strix then run the tests.
Odin Research is clear that this chain is an analytical model. Only a few servers ran two families side by side, and no single operator was seen running the whole chain. Still, the model explains why patch windows for internet-facing apps keep shrinking.
For defenders, the exposed surface is the issue. OWASP lists improper inventory management among the top API security risks. Forgotten API versions stay exposed and unpatched. AutoHunter-style recon finds them faster than most teams do.
What Odin Research observed
Monograph 20 covered a two-week window ending 18 November 2026.
- 21 endpoints on 21 hosts. Each AutoHunter install sat on its own server.
- A default port. 14 of the 21 (66.7%) used control port 18800. Single listeners sat on 80, 443, 8080 and several high ports.
- Hosting. China hosted 14 endpoints, 11 of them on Shenzhen Tencent. Hong Kong hosted 6, led by 4 on cognetcloud.
- One confirmed controller. A controller on cognetcloud in Hong Kong, seen on 13 November 2026, is confirmed and published for blocking. The other 20 are candidates and stay on watch.
- Neighbors. One Shenzhen Tencent host ran AutoHunter next to an AutoCVE console. A Hong Kong host paired it with a XingRin scanner node.
AutoHunter also shows up as background noise elsewhere. In Monograph 02, three hosts near Daybreak-labeled servers carried ARL or AutoHunter recon hits. Odin Research saw controller infrastructure only. Nothing in the data shows that any AutoHunter check reached or affected a target.
How to detect AutoHunter
- Outbound queue traffic. An internal host that posts to
/api/autohunter/with atarget_scopefield is running or feeding the agent. - Port 18800. Connections to 18800 from servers that have no testing role deserve a look. Pair the port with the route; do not act on the port alone.
- Inbound pattern. In your web logs, look for a crawl of every JavaScript bundle followed by systematic requests to each extracted route and parameter.
- Shared TLS fingerprints. Monograph 20 notes one JARM value shared by Strix, AutoHunter and XingRin listeners. Use it for hunting only, together with an API route.
Monograph 20 publishes a Suricata signature (SID 9520005) for AutoHunter queue dispatch and a YARA rule for Cairn and AutoHunter recon queue configuration files. Its KQL query flags contact with the confirmed controllers. All rules are on the report page and in the Defense Kit.
How to defend against AutoHunter-style recon
- Inventory your APIs. Keep a list of every API host, version and environment, as OWASP recommends. Retire versions you no longer need.
- Trim what the front end reveals. Do not ship routes for admin or internal functions in public JavaScript bundles.
- Rate-limit and alert. A web application firewall or API gateway can slow systematic parameter probing and flag it.
- Map your own surface first. External attack surface management shows you what an agent like AutoHunter would see.
- Allowlist your testers. Register approved scanner hosts before each test window, so real alerts stand out.
AutoHunter compared with other AI vulnerability agents
| Agent | Stage | Endpoints (Monograph 20) | Confirmed controllers |
|---|---|---|---|
| AutoHunter | Surface discovery: API routes and parameters | 21 | 1 |
| Cairn | Surface discovery: subdomains, certificates, services | 17 | 1 |
| AutoCVE | Turns advisories and patch diffs into HTTP checks | 118 | 7 |
| Strix | Runs web and API checks from Docker sandboxes | 40 | 5 |
| DeepAudit | Audits source code; dual-use | 61 | 0 |
Frequently asked questions about AutoHunter
What is AutoHunter?
AutoHunter is an automated reconnaissance agent for headless bug-bounty work. It extracts API routes and query parameters from a website's JavaScript and queues them for fuzzing.
Is AutoHunter malware?
No. It is a dual-use recon tool. Testers and bug-bounty hunters use tools like it legitimately. It becomes a threat when someone points it at systems they have no permission to test.
Is AutoHunter legal to use?
Running recon against systems you own or are authorized to test is legal. Scanning other organizations without permission can break computer misuse laws and bug-bounty program rules.
What port does AutoHunter use?
Odin Research saw 14 of 21 AutoHunter endpoints on control port 18800. Others used ports such as 80, 443 and 8080.
Who made AutoHunter?
We could not find an official public project page or a named author. Odin Research describes the tool from the servers it observed running it.
How do I detect AutoHunter?
Look for POST requests to /api/autohunter/ with a target_scope field, traffic to port 18800 from non-testing hosts, and systematic probing of every route in your JavaScript. Odin Research publishes a Suricata signature and a YARA rule in its Defense Kit.
AutoHunter vs Cairn: what is the difference?
Both do surface discovery. Cairn maps subdomains, certificates and services across a perimeter. AutoHunter maps the API routes and parameters inside individual web applications.
Sources
- MITRE ATT&CK, Active Scanning, T1595. attack.mitre.org/techniques/T1595/ (checked 21 November 2026)
- MITRE ATT&CK, Active Scanning: Wordlist Scanning, T1595.003. attack.mitre.org/techniques/T1595/003/ (checked 21 November 2026)
- OWASP API Security Top 10 2023, API9:2023 Improper Inventory Management. api-security.owasp.org/editions/2023/en/0xa9-improper-inventory-management (checked 21 November 2026)
Related tools and terms
What is CyberStrikeAI?
CyberStrikeAI is a public AI-driven offensive security platform. It spreads its operator interface, API and agent listener across several ports on one host and exposes tools through MCP.
AI offensive toolWhat is PentAGI?
PentAGI is an open-source autonomous penetration testing system that coordinates AI agents and security tools in containers.
AI offensive toolWhat is HexStrike AI?
HexStrikeAI is a public project that exposes many security tools to AI agents through the Model Context Protocol (MCP).
AI security toolWhat is DeepAudit?
DeepAudit is an open-source AI code auditing tool. Application security teams use it to find flaws in source code.
AI security toolWhat is AutoCVE?
AutoCVE is an AI-assisted tool that turns new vulnerability advisories into HTTP checks.
AI security toolWhat is Cairn?
Cairn is a reconnaissance tool that keeps an asset graph built from DNS zones, TLS certificates and HTTP banners.
Odin Research
Get the rules, not just the definition
Every Odin Research report ships with verified indicators and detection rules you can load today.