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
refactor(core): remove unreferenced helpers, attributes and a shadowed module (#4091)
Tejas Chopra08fce1831067
diff --git a/headroom/ccr/batch_processor.py b/headroom/ccr/batch_processor.pyindex bfe73b317..49abf4230 100644--- a/headroom/ccr/batch_processor.py+++ b/headroom/ccr/batch_processor.py@@ -23,7 +23,7 @@ from __future__ import annotationsimport jsonimport loggingfrom dataclasses import dataclass-from typing import Any, Protocol+from typing import Anyimport httpx@@ -34,19 +34,6 @@ from .tool_injection import CCR_TOOL_NAMElogger = logging.getLogger(__name__)-class APIClient(Protocol):- """Protocol for making API calls."""-- async def post(- self,- url: str,- headers: dict[str, str],- json: dict[str, Any],- ) -> httpx.Response:- """Make a POST request."""- ...--@dataclassclass BatchResultProcessorConfig:"""Configuration for batch result processing."""diff --git a/headroom/ccr/batch_store.py b/headroom/ccr/batch_store.pyindex 4bee79897..a68779a99 100644--- a/headroom/ccr/batch_store.py+++ b/headroom/ccr/batch_store.py@@ -126,7 +126,6 @@ class BatchContextStore:self._ttl = ttlself._max_contexts = max_contextsself._lock = asyncio.Lock()- self._cleanup_task: asyncio.Task | None = Noneasync def store(self, context: BatchContext) -> None:"""Store a batch context.diff --git a/headroom/ccr/mcp_server.py b/headroom/ccr/mcp_server.pyindex 53fe4f977..de0e0e045 100644--- a/headroom/ccr/mcp_server.py+++ b/headroom/ccr/mcp_server.py@@ -388,7 +388,6 @@ class HeadroomMCPServer:self._stats = SessionStats()self._local_store: Any = None # Lazy-initialized CompressionStoreself._token_estimator: EstimatingTokenCounter | None = None- self._compressor_initialized = False# File read cache: path → (content_hash, ccr_hash, line_count, token_count)self._file_cache: dict[str, tuple[str, str, int, int]] = {}diff --git a/headroom/cli.py b/headroom/cli.pydeleted file mode 100644index c7e8c4775..000000000--- a/headroom/cli.py+++ /dev/null@@ -1,6 +0,0 @@-"""Backwards compatibility - CLI moved to headroom.cli package."""--from headroom.cli import main--if __name__ == "__main__":- main()diff --git a/headroom/client.py b/headroom/client.pyindex c71133399..8ae403f10 100644--- a/headroom/client.py+++ b/headroom/client.py@@ -642,46 +642,6 @@ class HeadroomClient:self._storage.save(metrics)raise- def _call_openai(- self,- *,- model: str,- messages: list[dict[str, Any]],- stream: bool,- metrics: RequestMetrics,- **kwargs: Any,- ) -> Any:- """Call OpenAI-style API."""- return call_client_transport(- "openai",- self,- model=model,- messages=messages,- stream=stream,- metrics=metrics,- **kwargs,- )-- def _call_anthropic(- self,- *,- model: str,- messages: list[dict[str, Any]],- stream: bool,- metrics: RequestMetrics,- **kwargs: Any,- ) -> Any:- """Call Anthropic-style API."""- return call_client_transport(- "anthropic",- self,- model=model,- messages=messages,- stream=stream,- metrics=metrics,- **kwargs,- )-def _wrap_stream(self,stream: Iterator[Any],diff --git a/headroom/copilot_auth.py b/headroom/copilot_auth.pyindex d88bf27e4..a1fe3554e 100644--- a/headroom/copilot_auth.py+++ b/headroom/copilot_auth.py@@ -281,17 +281,6 @@ def _configured_api_url() -> str:return DEFAULT_API_URL-def copilot_api_url() -> str:- """Return the configured Copilot API base URL without any network calls.-- Resolves ``GITHUB_COPILOT_API_URL``, then the configured enterprise domain,- then ``api.githubcopilot.com``. Unlike :func:`resolve_copilot_api_url` this- performs no token exchange, so it is safe to call while routing a request.- """-- return _configured_api_url()--# GitHub's token exchange advertises the host that serves inline completions# under ``endpoints.proxy``, alongside the ``endpoints.api`` chat host. It is# recorded here when observed so completions routing uses GitHub's own answerdiff --git a/headroom/perf/analyzer.py b/headroom/perf/analyzer.pyindex 059cdd9ec..de32f4aca 100644--- a/headroom/perf/analyzer.py+++ b/headroom/perf/analyzer.py@@ -67,12 +67,11 @@ _STAGE_TIMINGS_RE = re.compile(# ----------------------------------------------------------------------------# Cache-aware pricing via LiteLLM+# List pricing via LiteLLM# ----------------------------------------------------------------------------# LiteLLM already knows per-token costs for 100+ models including-# cache_read and cache_creation pricing. We call it directly instead-# of maintaining our own pricing tables.+# LiteLLM already knows per-token costs for 100+ models. We read it+# directly instead of maintaining our own pricing tables.try:import litellm as _litellm@@ -82,32 +81,6 @@ except ImportError:_LITELLM_AVAILABLE = False-def _litellm_cost(- model: str,- prompt_tokens: int,- cache_read_tokens: int = 0,- cache_write_tokens: int = 0,-) -> float | None:- """Compute input cost via litellm.cost_per_token (cache-aware).-- Returns total input cost in USD, or None if model not found.- """- if not _LITELLM_AVAILABLE:- return None- resolved = resolve_litellm_model(model)- try:- input_cost, _ = _litellm.cost_per_token(- model=resolved,- prompt_tokens=prompt_tokens,- completion_tokens=0,- cache_read_input_tokens=cache_read_tokens,- cache_creation_input_tokens=cache_write_tokens,- )- return float(input_cost)- except Exception:- return None--def _get_list_price(model: str) -> float | None:"""Get list input price per 1M tokens."""if not _LITELLM_AVAILABLE:diff --git a/headroom/telemetry/models.py b/headroom/telemetry/models.pyindex 1c88499b6..86b25f44b 100644--- a/headroom/telemetry/models.py+++ b/headroom/telemetry/models.py@@ -633,38 +633,6 @@ class FieldSemantics:self.inferred_type = inferredself.confidence = confidence- def is_value_important(self, value_hash: str) -> bool:- """Check if a specific value is considered important.-- A value is important if:- 1. It's in the important_value_hashes list (has been retrieved)- 2. It's NOT the default value (for error_indicator type)-- Args:- value_hash: SHA256[:8] hash of the value to check.-- Returns:- True if this value should be preserved during compression.- """- # If we don't have enough data, be conservative- if self.confidence < 0.3:- return False-- # For error_indicator: non-default values are important- if self.inferred_type == "error_indicator":- if self.default_value_hash and value_hash != self.default_value_hash:- return True-- # For any type: values that have been retrieved are important- if value_hash in self.important_value_hashes:- return True-- # For status: check if this value has been retrieved- if self.inferred_type == "status":- return value_hash in self.value_retrieval_frequency-- return False-@dataclassclass CompressionEvent:diff --git a/headroom/telemetry/toin.py b/headroom/telemetry/toin.pyindex 5d68a900e..ccf62f2d1 100644--- a/headroom/telemetry/toin.py+++ b/headroom/telemetry/toin.py@@ -78,10 +78,6 @@ logger = logging.getLogger(__name__)# Environment variable for custom TOIN storage pathTOIN_PATH_ENV_VAR = "HEADROOM_TOIN_PATH"-# Default TOIN storage directory and file-DEFAULT_TOIN_DIR = ".headroom"-DEFAULT_TOIN_FILE = "toin.json"-# ── Aggregation-key defaults ────────────────────────────────────────────# Used when callers haven't plumbed auth-mode / model-family detection# (PR-F3 wires the real detectors). Shipping a real `"unknown"` slice isdiff --git a/pyproject.toml b/pyproject.tomlindex f51483723..92679bf00 100644--- a/pyproject.toml+++ b/pyproject.toml@@ -541,7 +541,6 @@ markers = [source = ["headroom"]branch = trueomit = [- "headroom/cli.py","*/tests/*",]