Flagged by the security agent. Hidden from the directory pending manual review.
Hosted VulKill MCP connector for authorized source-code and website security scans, account credits, and evidence-based reports.
# VulKill hosted security scanning
This connector provides access to the hosted VulKill MCP service at https://app.vulkill.com/api/mcp. It does not install the scanning engine in Cursor. Service tools and their definitions can change independently; review the current tool descriptions and requested permissions before connecting.
## Authorization and privacy
- Use the service only for assets the user owns or is explicitly authorized to test. For a live website, follow the service's target verification process.
- Before submitting a scan, explain that the selected source file contents or target URLs will be sent to the hosted service. Show the chosen scope and obtain the user's explicit permission for that transfer.
- Obtain the user's explicit acceptance of the current authorization and responsibility statement for the requested scan. Do not accept it on the user's behalf or infer consent from installation or sign-in.
- Do not include unrelated sensitive files. Obtain specific informed permission before including credentials or private data in an authorized security review. Never place access tokens or API keys in chat, source files, or this connector configuration.
- Scan reports can contain finding evidence, file locations, or code excerpts. Consult https://vulkill.com/privacy/ for processing, retention and user controls.
## Workflow
1. Connect the user's own VulKill account and inspect the current tool list, permissions, available credit and applicable request limits.
2. Ask which authorized files or verified HTTPS target the user wants to assess. Use only that scope and the appropriate scan mode.
3. Send selected source contents or the verified target to the hosted service only after the required consent. Respect file, request-size and credit limits; do not silently omit files or imply that a URL transfers local files.
4. Track the returned scan ID until the service reports its actual status. Do not treat a timeout, an accepted request or an estimate as a completed scan.
5. Summarize findings with their evidence, severity, limitations and available remediation guidance. Report completed, skipped and failed checks separately. Do not assert that a finding is exploitable or a patch is verified without supporting evidence.
## Security scope
The service advertises a catalog of more than 260 checks across 12 security areas. Applicability depends on the selected scan mode, the supplied files or target, and the capabilities actually run. The areas include:
1. Authentication, sessions and access control.
2. Business logic, payments and race conditions.
3. Code injection, command execution and unsafe data handling.
4. Network, SSRF and file-system security.
5. Secrets, credentials, Git-history indicators and cryptography within supplied content.
6. Supply chain, dependencies and software inventories.
7. AI, LLM and MCP agent security.
8. Web3, smart-contract source and cryptography, where supported.
9. Containers, cloud and infrastructure configuration.
10. Mobile application source and configuration relevant to APK/IPA, and desktop source such as Electron/Tauri, where supported.
11. Deep AST/data-flow, Z3 symbolic analysis and binary-hardening checks, where supported by the supplied input and engine.
12. Authorized website and API checks, three-layer findings triage, and remediation or patch guidance where generated.
Use the current VulKill capabilities and the scan's coverage report as the source of truth. A catalog entry is not evidence that a check ran, and an empty result is not a security guarantee.
Service: https://vulkill.com/ | Privacy: https://vulkill.com/privacy/ | Support: https://vulkill.com/#iletisim
Do not assume that the remote service can read a local path or private repository: use only a currently supported transfer or integration that the user has authorized.