# Sabrix-bench: Fast MCP inspector and agent loop latency profiler in Rust

> Source: <https://github.com/Pro-Kla/sabrix-bench>
> Published: 2026-08-19 19:55:41+00:00

Ultra-fast, zero-bloat developer CLI and benchmark harness for Model Context Protocol (MCP) traffic inspection and agent loop latency profiling.

Developers building autonomous AI agents locally lack visibility into raw **Model Context Protocol (MCP)** JSON-RPC tool traffic and have no lightweight way to measure per-turn serialization and proxy latency overhead.

Legacy approaches introduce massive performance and security taxes:

**Legacy Python / Node Proxies:** Add**30ms – 50ms** of serialization and runtime tax per tool call.**SaaS AI Firewalls:** Incur**100ms – 250ms** of WAN network latency, TLS handshakes, and third-party cloud data egress per turn.- Over a
**30-turn agent loop**, legacy firewalls add** 3.6+ seconds of dead wait time**and leak raw database queries and internal system commands outside your perimeter.

`sabrix-bench`

gives you real-time visibility into your local MCP tool calls and benchmarks your agent loops in **microseconds ( $< 2\ \mu\text{s}$)** with

**zero network egress**.

```
# Clone the repository and install the binary
cargo install --path .
```

Or run directly with Cargo:

```
cargo run -- --help
```

Inspect JSON-RPC 2.0 requests (`tools/call`

, `resources/read`

, `tools/list`

) and flag dangerous tool mutations (destructive shell commands, SQL injections, leaked credentials).

```
sabrix-bench trace --demo
sabrix-bench trace -p '{
  "jsonrpc": "2.0",
  "id": 1,
  "method": "tools/call",
  "params": {
    "name": "database_query",
    "arguments": { "sql": "DROP TABLE users; --" }
  }
}'
cat mcp_message.json | sabrix-bench trace
```

Simulate multi-turn autonomous agent loops to measure local parsing overhead, memory serialization, and latency percentiles (

```
# Run a 30-turn benchmark with real-time progress
sabrix-bench bench --turns 30

# Benchmark larger context payloads (10x scale)
sabrix-bench bench --turns 50 --scale 10

# Output machine-readable JSON for CI/CD pipelines
sabrix-bench bench --turns 20 --json
```

Output a side-by-side architectural comparison contrasting In-VPC embedded engines against legacy proxies and remote SaaS firewalls.

```
sabrix-bench compare
```

| Layer / Architecture | Per-Turn Latency | 30-Turn Loop Delay | Egress & Privacy | Memory Footprint |
|---|---|---|---|---|
Sabrix In-VPC Engine |
(0.002 ms)`< 2 µs` |
`< 0.06 ms` |
100% In-VPC (Zero Egress) |
`< 15 MB` |
Legacy Python / Node Proxy |
`35.0 ms` |
`+1.05 seconds` |
Local Cluster | `150 MB – 400 MB` |
SaaS AI Firewall |
`120.0 ms` |
`+3.60 seconds` |
Full Payload Egress | N/A (Cloud SaaS) |

`sabrix-bench`

evaluates deterministic security rules locally in sub-microsecond time:

: Destructive Filesystem Operations (`MCP-SEC-001`

`rm -rf`

,`mkfs`

,`dd`

,`chmod 777`

): Remote Code Execution / Reverse Shells (`MCP-SEC-002`

`curl | bash`

,`nc -e`

,`/dev/tcp/`

): Destructive SQL Mutations & Injections (`MCP-SEC-003`

`DROP TABLE`

,`TRUNCATE`

,`DELETE FROM`

,`WHERE 1=1`

): Leaked API Keys & Secrets (OpenAI`MCP-SEC-004-007`

`sk-`

, GitHub`ghp_`

, AWS`AKIA`

, PEM Private Keys): Sensitive Path Egress (`MCP-SEC-008`

`/etc/passwd`

,`~/.ssh/id_rsa`

,`~/.aws/credentials`

,`.env`

): Unconstrained Arbitrary Execution Tool Invocations`MCP-SEC-009`

Enforce zero-egress In-VPC MCP security with millisecond-grade deterministic policy control:

👉 [Deploy Sabrix In-VPC Gateway](https://sabrix.ai)

Licensed under either of [Apache License, Version 2.0](/Pro-Kla/sabrix-bench/blob/main/LICENSE-APACHE) or [MIT license](/Pro-Kla/sabrix-bench/blob/main/LICENSE-MIT) at your option.