Configuration

Configuration Reference

Overview

M2M Protocol can be configured via:

Configuration file (~/.m2m/config.toml)
Environment variables
Command-line arguments

Priority: CLI > Environment > Config File > Defaults

Configuration File

Default location: ~/.m2m/config.toml

Complete Example

# M2M Protocol Configuration

[proxy]
# Address to listen on
listen = "127.0.0.1:8080"

# Upstream LLM API endpoint
upstream = "http://localhost:11434/v1"

# API key for upstream (optional)
api_key = ""

# Request timeout in seconds
timeout = 30

[proxy.compression]
# Compress outgoing requests
requests = true

# Compress incoming responses
responses = true

# Preferred algorithm (token, brotli, auto)
algorithm = "auto"

[security]
# Enable security scanning
enabled = true

# Block unsafe content (vs just logging)
blocking = true

# Confidence threshold (0.0 - 1.0)
threshold = 0.8

[compression]
# Use ML model for algorithm routing
ml_routing = false

# Minimum size for Brotli (bytes)
brotli_threshold = 4096

# Prefer token compression for API payloads
prefer_token_for_api = true

[logging]
# Log level: trace, debug, info, warn, error
level = "info"

# JSON format for structured logging
json = false

Environment Variables

Variable	Description	Default
`M2M_SERVER_PORT`	Proxy listen port	`8080`
`M2M_SERVER_HOST`	Proxy listen address	`127.0.0.1`
`M2M_UPSTREAM_URL`	Upstream API URL	-
`M2M_API_KEY`	API key for upstream	-
`M2M_SECURITY_ENABLED`	Enable security scanning	`true`
`M2M_SECURITY_BLOCKING`	Block unsafe content	`true`
`M2M_BLOCK_THRESHOLD`	Security confidence threshold	`0.8`
`M2M_LOG_LEVEL`	Logging level	`info`
`M2M_LOG_JSON`	JSON log format	`false`
`M2M_TIMEOUT`	Request timeout (seconds)	`30`

CLI Arguments

Server Command

m2m server [OPTIONS]

Options:
  -p, --port <PORT>          Listen port [default: 8080]
  -h, --host <HOST>          Listen address [default: 127.0.0.1]
  -u, --upstream <URL>       Upstream API URL [required]
  -k, --api-key <KEY>        API key for upstream
  --security                  Enable security scanning
  --threshold <FLOAT>        Security threshold [default: 0.8]
  --timeout <SECONDS>        Request timeout [default: 30]
  --log-level <LEVEL>        Log level [default: info]
  --log-json                 JSON log format

Compress Command

m2m compress [OPTIONS] <CONTENT>

Arguments:
  <CONTENT>                  Content to compress (or - for stdin)

Options:
  -a, --algorithm <ALG>      Algorithm (token, brotli, auto) [default: auto]
  -o, --output <FILE>        Output file (default: stdout)

Decompress Command

m2m decompress [OPTIONS] <CONTENT>

Arguments:
  <CONTENT>                  Content to decompress (or - for stdin)

Options:
  -o, --output <FILE>        Output file (default: stdout)

Scan Command

m2m scan [OPTIONS] <CONTENT>

Arguments:
  <CONTENT>                  Content to scan

Options:
  -t, --threshold <FLOAT>    Confidence threshold [default: 0.8]
  --json                     JSON output format

Proxy Configuration Details

Listen Address

[proxy]
listen = "127.0.0.1:8080"  # Local only
listen = "0.0.0.0:8080"    # All interfaces (caution!)

Upstream Configuration

[proxy]
# Local Ollama
upstream = "http://localhost:11434/v1"

# OpenAI
upstream = "https://api.openai.com/v1"
api_key = "sk-..."

# Azure OpenAI
upstream = "https://YOUR-RESOURCE.openai.azure.com/openai/deployments/YOUR-DEPLOYMENT"
api_key = "..."

Security Configuration

Scanning Modes

Mode	`enabled`	`blocking`	Behavior
Disabled	false	-	No scanning
Monitor	true	false	Scan and log, allow all
Blocking	true	true	Scan and reject threats

Threshold Tuning

Threshold	False Positives	False Negatives
0.5	High	Low
0.8	Medium	Medium
0.95	Low	High

Compression Configuration

Algorithm Selection

Setting	Description
`auto`	Automatically select best algorithm
`token`	Always use token compression
`brotli`	Always use Brotli compression
`none`	Disable compression

Size Thresholds

[compression]
# Minimum content size for compression (bytes)
min_size = 100

# Use Brotli above this size (bytes)
brotli_threshold = 4096

# Maximum content size (bytes)
max_size = 16777216

Logging Configuration

Log Levels

Level	Description
`trace`	Very detailed debugging
`debug`	Debugging information
`info`	Normal operation
`warn`	Warning conditions
`error`	Error conditions

JSON Logging

[logging]
json = true

Output:

{"timestamp":"2026-01-17T12:00:00Z","level":"INFO","message":"Request compressed","algorithm":"token","ratio":0.66}