T05 · Unauthorized Access and Privilege Escalation
- Location
healthprobe/app.py:27- Finding
Unrestricted Outbound Requests Enable Server-Side Request Forgery
- Content
View full analysis
- Remediation
View remediation
Security audit
Security checks for vulnerabilities and agentic risk
HealthProbe does what it says, but it exposes an unrestricted URL-probing service that could be abused to reach internal network resources from the host running it.
Install only if you are comfortable running a local URL-probing service from your environment. Avoid exposing it beyond localhost, do not use it in shared or untrusted settings without authentication and network egress controls, and consider blocking private, loopback, link-local, and cloud metadata destinations before use.
healthprobe/app.py:27Unrestricted Outbound Requests Enable Server-Side Request Forgery
healthprobe/app.py:37Raw Network Exception Messages Are Returned to Untrusted Callers
SKILL.md:4Runtime Dependencies Are Installed Without Version or Integrity Pinning
The skill explicitly performs outbound network activity ('probe any URL') yet does not declare any tool scope or permissions boundaries. This weakens reviewability and policy enforcement, and in practice can enable unintended network access patterns such as probing internal or sensitive endpoints if the runtime does not impose additional controls.
Data is being sent to an external URL. This could be legitimate telemetry or data exfiltration. Manual review is recommended.
curl -s -X POST http://localhost:8009/v1/probe \
-H "Content-Type: application/json" \
-d '{"url": "https://example.com"}' | jq
Data is being sent to an external URL. This could be legitimate telemetry or data exfiltration. Manual review is recommended.
curl -s -X POST http://localhost:8009/v1/probe \
-H "Content-Type: application/json" \
-d '{"url": "https://example.com"}' | jq
The endpoint performs arbitrary outbound HTTP requests to a user-supplied URL with redirects enabled, effectively creating a network probing primitive. In an agent or multi-tenant environment, this can be abused for SSRF-style access to internal services, cloud metadata endpoints, or network reconnaissance, and the lack of any warning or restriction increases the risk of misuse.