Compress tool outputs, logs, files, and RAG chunks before they reach the LLM. 20% fewer tokens for coding agents, 60-95% fewer tokens for JSON, same answers. Library, proxy, MCP server.
Python agentaianthropicclaude-codecompressioncontext-engineeringcontext-windowcursorfastapilangchainllmmcpopenaiprompt-engineeringproxypythonragtoken-optimizationtokenstypescript
"""Tests for CO₂ saved estimation in savings_tracker.
Validates estimate_co2_saved_mg() and its presence in stats_preview().
"""
from __future__ import annotations
from headroom.proxy.savings_tracker import estimate_co2_saved_mg
def test_zero_tokens_returns_zero() -> None:
assert estimate_co2_saved_mg(0) == 0.0
assert estimate_co2_saved_mg(-100) == 0.0
def test_positive_tokens_returns_positive() -> None:
result = estimate_co2_saved_mg(1_000_000)
assert result > 0
def test_claude_sonnet_uses_correct_factor() -> None:
# claude-sonnet: 0.0003 Wh/token × 0.4 kg/kWh × 1000 mg/g = 0.12 mg/token
result = estimate_co2_saved_mg(1000, model="claude-sonnet-4-5")
assert abs(result - 120.0) < 1.0 # 0.12 mg × 1000 = 120 mg
def test_unknown_model_uses_default_factor() -> None:
# default: 0.0002 Wh/token × 0.4 = 0.08 mg/token
result = estimate_co2_saved_mg(1000, model="unknown-model-xyz")
assert abs(result - 80.0) < 1.0
def test_claude_haiku_lower_than_sonnet() -> None:
haiku = estimate_co2_saved_mg(1000, model="claude-haiku")
sonnet = estimate_co2_saved_mg(1000, model="claude-sonnet")
assert haiku < sonnet
def test_large_savings_in_grams() -> None:
# 1M tokens at claude-sonnet = 120,000 mg = 120 g
mg = estimate_co2_saved_mg(1_000_000, model="claude-sonnet")
g = mg / 1000
assert abs(g - 120.0) < 1.0
def test_stats_preview_includes_co2() -> None:
"""stats_preview() must include a co2 block with co2_saved_mg."""
import pathlib
import tempfile
from headroom.proxy.savings_tracker import SavingsTracker
with tempfile.TemporaryDirectory() as tmp:
tracker = SavingsTracker(path=str(pathlib.Path(tmp) / "savings.json"))
tracker.record_compression_savings(
model="claude-sonnet-4-5",
tokens_saved=10_000,
)
preview = tracker.stats_preview(model="claude-sonnet-4-5")
assert "co2" in preview
co2 = preview["co2"]
assert "co2_saved_mg" in co2
assert co2["co2_saved_mg"] > 0
assert "co2_saved_g" in co2
assert co2["co2_saved_g"] == round(co2["co2_saved_mg"] / 1000, 6)
assert "methodology" in co2