JevGuard

# JevGuard: Deterministic Caching and Guardrails for TypeSafe AI (Jev)

[](https://pypi.org/project/jevguard-core/)
[](https://pypi.org/project/jevguard-mcp/)
[](https://www.python.org/)
[](LICENSE)

> **Official Model Context Protocol (MCP) Server:** JevGuard includes an official, zero-dependency MCP server designed for autonomous AI agents in Google Antigravity, Cursor IDE, Claude Desktop, and Cline: **[seb4ez/jevguard-mcp](https://github.com/seb4ez/jevguard-mcp)**.

JevGuard is an open-source Python library that provides local deterministic caching, state pruning, neutral escape injection, and calibration guardrails around TypeSafe AI's System One decision model. While upstream decision engines produce probabilistic outputs, JevGuard provides a strictly deterministic local layer: canonical JSON sorting, SHA-256 fingerprinting with volatile key masking, SQLite storage, and rule-based calibration checks using only the Python standard library.

## Upstream Latency and Cache Performance (50 Live Verification Requests)

The following benchmark report reflects 50 complete test requests executed against the official TypeSafe AI endpoint (`https://api.typesafe.ai/v1/systemone` using model `jev-latest`) from a development workstation. Latency comparisons contrast WAN roundtrips against local in-memory/SQLite cache lookups:



| Key Metric | Direct Upstream API | JevGuard Runtime | Empirical Advantage |
| :--- | :--- | :--- | :--- |
| **Verification Requests** | 50 live calls tested | 50 live calls tested | 100% transport pass rate across 5 test scenarios |
| **Average Network Latency** | 784.53 ms | 735.88 ms (pruned) | State pruning reduces wire payload size |
| **Deterministic Cache Hit** | 763.40 ms (144 tokens charged) | **0.099 ms** (**0 tokens**) | **Sub-millisecond ret