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 pure proxy semantic-cache key policy."""
from __future__ import annotations
from headroom.proxy.semantic_cache_key_policy import (
compute_semantic_cache_key,
strip_cache_control,
)
MESSAGES = [{"role": "user", "content": "hello"}]
MODEL = "claude-haiku-4-5"
def test_semantic_cache_key_distinguishes_response_shaping_fields() -> None:
assert compute_semantic_cache_key(MESSAGES, MODEL, partition="p_test", system="French") != (
compute_semantic_cache_key(MESSAGES, MODEL, partition="p_test", system="English")
)
assert compute_semantic_cache_key(MESSAGES, MODEL, partition="p_test", temperature=0.0) != (
compute_semantic_cache_key(MESSAGES, MODEL, partition="p_test", temperature=1.0)
)
def test_semantic_cache_key_ignores_moved_cache_control() -> None:
with_cache_control = [{"type": "text", "text": "sys", "cache_control": {"type": "ephemeral"}}]
without_cache_control = [{"type": "text", "text": "sys"}]
assert compute_semantic_cache_key(
MESSAGES, MODEL, partition="p_test", system=with_cache_control
) == (
compute_semantic_cache_key(
MESSAGES, MODEL, partition="p_test", system=without_cache_control
)
)
def test_semantic_cache_key_ignores_moved_message_cache_control() -> None:
"""cache_control on a message must not fragment the key either (#327 intent)."""
messages_with_cc = [
{
"role": "user",
"content": [{"type": "text", "text": "hello", "cache_control": {"type": "ephemeral"}}],
}
]
messages_without_cc = [{"role": "user", "content": [{"type": "text", "text": "hello"}]}]
assert compute_semantic_cache_key(messages_with_cc, MODEL, partition="p_test") == (
compute_semantic_cache_key(messages_without_cc, MODEL, partition="p_test")
)
def test_strip_cache_control_recurses_through_dicts_and_lists() -> None:
assert strip_cache_control(
{
"system": [
{"type": "text", "text": "sys", "cache_control": {"type": "ephemeral"}},
],
"tools": [{"name": "read", "cache_control": {"type": "ephemeral"}}],
}
) == {
"system": [{"type": "text", "text": "sys"}],
"tools": [{"name": "read"}],
}
def test_strip_cache_control_preserves_user_defined_schema_property() -> None:
schema = {
"type": "object",
"properties": {
"cache_control": {
"type": "boolean",
"description": "Flush the application cache",
}
},
}
assert strip_cache_control(schema) == schema