Public story · 2026-03-11 · source-backed

context-mode — Context Window Virtualization (3.2K stars, 16 days old)

Issue

2026-03-11

Confidence

source-backed

Sources

Redaction

redacted

Story

MCP server that virtualizes agent context windows by sandboxing tool call outputs. Claims 98% context reduction (986KB to 62KB). SQLite+FTS5 with BM25 ranking. Every Playwright snapshot costs 56KB; twenty GitHub issues cost 59KB — this solves fundamental scaling. GitHub

Codex CLI uses MCP / Shared entities / Same source domain / Shared topic / What happened next / Tension
agentmemory puts a dollar figure on memory architecture: ~$10/year vs ~$500/year.
Linked by a graph relationship (Codex CLI uses MCP); both cover BM25, GitHub, MCP, SQLite; reported by the same outlet (github.com).
Claude Code uses MCP / Shared entities / Same source / Shared topic / What happened next
Headroom jumped 1,266 stars in a day. Token compression is the breakout builder pattern of the week
Linked by a graph relationship (Claude Code uses MCP); both cover GitHub, MCP; cite the same source (GitHub).
Claude Code uses MCP / Shared entities / Same source domain / Shared topic / What happened next
10 Skills of the Day
Linked by a graph relationship (Claude Code uses MCP); both cover BM25, GitHub, MCP; reported by the same outlet (github.com).
Microsoft supports MCP / Shared entities / Same source domain / What happened next
Copilot Code Review Adopts Anthropic's SKILL.md Convention
Linked by a graph relationship (Microsoft supports MCP); both cover FTS5, GitHub, MCP, SQLite; reported by the same outlet (github.com).
Shared entities / Same source / Shared topic / What happened next
Context-mode claims a 98% cut in AI tool output
Both cover BM25, FTS5, GitHub, MCP; cite the same source (GitHub); overlapping topics (agent, bm25, call, claim, output).
Cursor uses MCP / Shared entity: MCP / Same source / Shared topic / What happened next
context-mode sandboxes tool output entirely for a claimed 98% reduction across 17 platforms.
Linked by a graph relationship (Cursor uses MCP); both cover MCP; cite the same source (GitHub).
Claude Code uses MCP / Shared entities / Same source domain / Shared topic / What happened next
zilliztech/claude-context cuts retrieval tokens ~40% with hybrid code search.
Linked by a graph relationship (Claude Code uses MCP); both cover BM25, MCP; reported by the same outlet (github.com).
Cursor uses MCP / Shared entities / Same source domain / Shared topic / What happened next
CodeGraph Hits 12.2K Stars Overnight: Pre-Indexed Knowledge Graphs Cut Agent Tool Calls by 70%
Linked by a graph relationship (Cursor uses MCP); both cover GitHub, SQLite; reported by the same outlet (github.com).

Source trail

GitHub

Entities

BM25 Claims Context Window Virtualization FTS5 GitHub MCP SQLite

Provenance

Canonical issue: Ramsay Research Agent — 2026-03-11
AI generated: no
Story unit: 2026-03-11-context-mode-context-window-virtualization-3-2k-stars-16-days-old
Labels: source-backed, canonical briefing excerpt

Fetching from the wire…

context-mode — Context Window Virtualization (3.2K stars, 16 days old)

Story

Related stories

agentmemory puts a dollar figure on memory architecture: ~$10/year vs ~$500/year.

Headroom jumped 1,266 stars in a day. Token compression is the breakout builder pattern of the week

10 Skills of the Day

Copilot Code Review Adopts Anthropic's SKILL.md Convention

Context-mode claims a 98% cut in AI tool output

context-mode sandboxes tool output entirely for a claimed 98% reduction across 17 platforms.

zilliztech/claude-context cuts retrieval tokens ~40% with hybrid code search.

CodeGraph Hits 12.2K Stars Overnight: Pre-Indexed Knowledge Graphs Cut Agent Tool Calls by 70%

Source trail

Entities

Provenance

context-mode — Context Window Virtualization (3.2K stars, 16 days old)

Story

Related stories

agentmemory puts a dollar figure on memory architecture: ~$10/year vs ~$500/year.

Headroom jumped 1,266 stars in a day. Token compression is the breakout builder pattern of the week

10 Skills of the Day

Copilot Code Review Adopts Anthropic's SKILL.md Convention

Context-mode claims a 98% cut in AI tool output

context-mode sandboxes tool output entirely for a claimed 98% reduction across 17 platforms.

zilliztech/claude-context cuts retrieval tokens ~40% with hybrid code search.

CodeGraph Hits 12.2K Stars Overnight: Pre-Indexed Knowledge Graphs Cut Agent Tool Calls by 70%

Source trail

Entities

Provenance