Compact Data

Reduce token count from CLI or Python API.

CLI

From stdin

Bash

echo '{"name": "Alice", "age": 30}' | axm-smelt compact

The compacted text goes to stdout. Savings are reported to stderr:

Text Only

{"name":"Alice","age":30}
Tokens: 14 -> 9 (35.7% saved)

From a file

Bash

axm-smelt compact --file data.json

Write output to a file

Bash

axm-smelt compact --file data.json --output compacted.json

The savings line is still printed to stderr; stdout is not used when --output is set.

Choose a preset

Bash

# Lossless only
axm-smelt compact --file data.json --preset safe

# Structural transforms
axm-smelt compact --file data.json --preset moderate

# Maximum savings
axm-smelt compact --file data.json --preset aggressive

Choose specific strategies

Bash

# Comma-separated strategy names
axm-smelt compact --file data.json --strategies minify,drop_nulls

Python API

Default preset (safe)

Python

from axm_smelt import smelt

report = smelt('{\n  "name": "Alice",\n  "age": 30\n}')
print(report.compacted)        # {"name":"Alice","age":30}
print(report.savings_pct)      # 35.71...
print(report.original_tokens)  # 14
print(report.compacted_tokens) # 9

With a preset

Python

report = smelt(data, preset="moderate")
print(report.strategies_applied)  # ['minify', 'drop_nulls', ...]

With explicit strategies

Python

report = smelt(data, strategies=["minify", "drop_nulls"])

SmeltReport fields

Field	Type	Description
original	str	Input text
compacted	str	Compacted text
original_tokens	int	Token count before
compacted_tokens	int	Token count after
savings_pct	float	Percentage saved
format	Format	Detected input format
strategies_applied	list[str]	Strategies that changed the output
strategy_estimates	dict[str, float]	Per-strategy savings estimates (from check only)