🛡️ correctover-runtime-guard
Real-time RCE / SSRF / credential-leak interception for AI Agent tool calls — sub-millisecond, one full pass over every pattern.
Quick Start
pip install correctover-runtime-guard
correctover-runtime-guard diagnose "run eval(x) and curl http://169.254.169.254/"
What It Does
Runtime guardrail for AI Agent tool calls — it inspects input before execution and blocks the fault classes that actually appear in MCP/agent incident reports:
- RCE interception (CWE-78) — shell command execution, destructive commands, fork bombs
- SSRF blocking (CWE-918) — HTTP client calls, cloud metadata endpoints / internal IPs
- Credential-leak prevention (CWE-200) — environment access and secret/API-key patterns
- Injection detection (CWE-89/CWE-78) — SQL and script-injection patterns
- MCP Server mode —
diagnose_error/get_fault_pattern/get_repair_suggestion
Each finding maps to a public CWE/CVE threat reference (CVE + CVSS shown only where a real, NVD-verifiable advisory exists). This is local pattern/taxonomy detection; it is not a claim of a measured detection rate against any benchmark.
Performance
Every call runs one complete pass over all 8 patterns — it never stops at the first hit, so the number is stable whether the input is clean or malicious.
| Metric (steady state, warm, local) | Value |
|---|---|
End-to-end scan() P50 |
~16 µs |
End-to-end scan() P99 |
~26 µs |
| Regex hot-path-only P50 / P99 | ~14 / ~24 µs |
Methodology: Python 3.13, single thread, 50,000 consecutive scans over a fixed input, after warm-up; P50/P99 taken over the run, with the first and last 1,000 samples matching (no latency growth with uptime). The first call after process start pays a one-time regex compilation cost (cold start) and is excluded. These are indicative local measurements, not a vendor benchmark — numbers vary with hardware and workload, so benchmark on your own target environment.
Free Tier
50 scans/day — no credit card required. Pro and Enterprise tiers add full diagnosis, repair strategies, live MCP interception and auto-heal. Pricing and activation:
export CORRECTOVER_LICENSE_KEY=your-key-here
Related Correctover Tools
| Tool | Install | Description |
|---|---|---|
| Security Scanner (JS/TS) | npx correctover-scan |
CCS audit for MCP configs and published JS/npm bundles |
| Runtime SDK (JS/TS) | npm install correctover |
Runtime verification SDK (CCS 7-dimension, Ed25519 receipts) |
| Self-Healing Test | pip install correctover-test |
Agent self-healing test suite with failover verification |
| Security Audit | pip install correctover-security-audit |
MCP/agent framework audit powered by CCS fault taxonomy v2.5 |
| Compliance Check | pip install correctover-compliance-check |
OAuth 2.1 + CCS + MCP conformance checks |
Website: correctover.com
1 maintainer
Metadata
1 maintainer
Release files for correctover-runtime-guard 2.1.5
For a detailed explanation of source distributions (sdists) and built distributions (wheels), please see the package formats documentation.
Built distribution (wheel)
| File | Interpreter | ABI | Platform | Reset |
|---|---|---|---|---|
| correctover_runtime_guard-2.1.5-py3-none-any.whl | Python 3 | none | any | Details |
Release files / correctover_runtime_guard-2.1.5-py3-none-any.whl
| Download URL | correctover_runtime_guard-2.1.5-py3-none-any.whl |
|---|---|
| Size | 22.0 kB |
| Tags | Python 3 |
|
SHA-256 checksum How to use checksums |
9ed7efe568225fd8c0c467433eb9334340d4e1ac163f75df9f3464473e7971d3
|
|
BLAKE2b-256 checksum How to use checksums |
4c1b76295ba2bd98877d7860bcbe592b7c2d9ce7d8c53f554090d170eb37c05c
|
| Upload date | |
|
Uploaded using Trusted Publishing? What is trusted publishing? |
No |
| Uploaded via |
twine/7.0.0 CPython/3.13.12
|
Security for correctover-runtime-guard 2.1.5
Provenance
Provenance describes where a file came from. On PyPI, provenance is shared via attestations, which provide a verifiable record of the build or publishing details. View details, limitations and caveats.
No provenance found
None of the files in this release were uploaded with provenance metadata.
A note on package security
PyPI is a repository for distributing packages, not a service for auditing them. While third-party security researchers and our staff do their best to detect and remove malicious packages, PyPI cannot guarantee the quality, security or suitability of any given package.
