ContextOS: Open-source token-budget-aware context orchestration for long-horizon agents
Whyrureadingthisz · reddit · 2026-08-23
The author released ContextOS, an open-source, token-budget-aware context orchestration layer for long-horizon LLM agents. It distinguishes between retrieval and context selection, utilizing hybrid retrieval (dense + BM25), RRF fusion, cross-encoder reranking, and deterministic planning to manage context budgets. The system logs execution traces for inspecting selection decisions, rankings, and token consumption. An evaluation harness and interactive demo are provided.
More from coding & agent
- LangChain answers the most common questions about agent evals: datasets, judges, and trajectories — LangChain · 2026-10-16
- Signal65 benchmarks: Intel Xeon 6 runs agent workloads 11% faster for less money — ryanshrout · 2026-10-16
- Vercel Daemons in production: log delivery pipeline hits 1 Gbit per second and ramping — cramforce · 2026-10-16
- From Mathematica to ~1000 Python scripts: a 20-year math visualization journey — S_Conradi · 2026-10-16
- How to Get Better at Delegating to Your AI Assistant: It's a Learnable Skill — craigsdennis · 2026-10-16
- HeyGen MCP turns ChatGPT recipes into personalized cooking videos — toolstelegraph · 2026-10-16