Restore pythonPath and proxyPort config support

Clarify local npm install path for launcher detection

Auto-detect proxyUrl by default and keep explicit override

Add ordered launcher fallback for headroom proxy startup

Add controlled proxy autostart and unsafe-install guidance

Harden openclaw plugin proxy handling and metadata

Bump to 0.5.18

Fix image token estimate: use 1600 tokens (pixel-based), not 85

Fix image token counting: Anthropic images counted as text tokens

Fix WebSocket SSL on Windows: use native websockets SSL handling

Disable CacheAligner: it inflates tokens and breaks prefix caching

Fix Codex WS proxy (Issue #86), LiteLLM metrics, and backend forwarding

Refactor: extract data models to headroom/proxy/models.py (Step 1/9)

Add transforms_summary to TransformResult and /v1/compress response

Merge pull request #81 from dorphalsig/main

Fix multi-worker beacon spam, cost dashboard, upsert telemetry; bump 0.5.17

Fix dashboard cost savings: use model list price, not moving average

Merge pull request #61 from gglucass/feat/persist-savings-history

Fix WebSocket proxy: forward all headers + required OpenAI-Beta header

Add Anthropic url overrides via ANTHROPIC_TARGET_URL env var and --anthropic-api-url flag

Update dashboard.html

Merge remote-tracking branch 'origin/main' into feat/persist-savings-history

style: format savings history tests

Update test_proxy_savings_history.py

feat: add historical rollups and exports

Bump to 0.5.16: WebSocket proxy, Kompress ONNX, Discord link update

Kompress ONNX INT8: text compression without torch dependency

Update Discord invite link across all docs

Add WebSocket proxy for /v1/responses (Codex gpt-5.4+)

Remove dedupe_telemetry.py from repo, add scripts/ to .gitignore

Bump to 0.5.15: Responses API compression, tool_result ordering fix

Add compression for OpenAI Responses API (/v1/responses)

Fix tool_result ordering: no stray user message between tool_calls and tool

Fix zstd streaming decompression, add zstandard dep; bump to 0.5.14

Fix streaming tool calls, compressed request bodies, beacon field names; bump to 0.5.13

Switch telemetry to proxy_telemetry_v2 table, bump to 0.5.12

Bump to 0.5.11: fix cost calculation, Windows path decoding, unused imports

Fix cost calculation: don't mix cache savings into Headroom savings

fix: remove unused imports in test_langgraph.py (ruff F401)

Fix Windows drive letter path decoding in headroom learn (fixes #69)

Merge branch 'main' into feat/persist-savings-history

test: cover savings tracker branches

Update savings_tracker.py

fix: restore anthropic compression when hooks enabled

refactor: remove dashboard storage path label

refactor: move dashboard mode switch to header

Merge pull request #60 from gglucass/feat/add-headroom-sdk

Merge pull request #68 from KunalLohtia/feat/langgraph-compress-tool-messages

feat: add LangGraph compress_tool_messages node for ToolMessage compression

docs: add OpenClaw plugin to README, integration guide, index, and TS SDK docs