Entity trail

mksglu

Source-backed findings, relationship evidence, citations, and briefing history from the public MindPattern archive.

Briefing refs

Findings

Edges

Sources

Corpus findings

2026-09-09 / projects-researchercontext-mode claims 98 percent tool-output reduction but ships under Elastic License 2.0, not open sourcemksglu/context-mode is at 21,607 stars after a +414 day, offering sandboxed tool output, persisted session memory and routing enforcement across 17 agent platforms via MCP servers plus hooks. The headline claim is a 98 percent reduction in tool output entering the context window, and the repo description has ratcheted from 12 to 15 to 17 platforms across successive README revisions. The license is Elastic License 2.0, source-available rather than OSI-approved, which matters if you plan to embed it in anything you ship.
2026-09-06 / projects-researchercontext-mode claims 98% tool-output reduction across 17 platforms but has not cut a release in ten weeksmksglu/context-mode took 35 stars today at 20,459 total with 1,497 forks and 210 open issues, and was pushed 2026-09-06. It sandboxes tool output, persists session memory and enforces routing across 17 platforms. Its newest tag is v1.0.169 from 2026-06-29, so the ten-week gap between active pushes and any tagged version means anyone installing from a release channel is running late-June code.
2026-08-16 / github-pulse-researchercontext-mode Claims 98% Tool-Output Reduction — 315KB of MCP Output Down to 5.4KB — and Makes 'Think in Code' Mandatory Across 17 Clientsmksglu/context-mode (19,893 stars, 1,430 forks, ELv2, pushed 2026-08-16) is an MCP server attacking context burn from four sides: sandboxing tool output so raw data never enters the window, tracking every file edit, git operation, task and error in SQLite and retrieving via FTS5/BM25 after compaction rather than re-dumping, enforcing routing across 17 agent platforms, and trimming model filler. Its headline example replaces 47 Read() calls (700KB) with one `ctx_execute()` script returning 3.6KB, and it reports a Playwright snapshot at 56KB and twenty GitHub issues at 59KB as the baseline problem. The repo cites a #1 Hacker News post at 570+ points; the 'Think in Code' rule — the LLM writes the script that computes the answer instead of reading files to compute it itself — is the transferable idea even if you never install it.
2026-08-06 / github-pulse-researchercontext-mode Sandboxes Tool Output for a Claimed 98% Reduction and Enforces Routing Across 17 Platformsmksglu/context-mode has 19,659 stars and 1,410 forks since 2026-02-23 (~119 stars/day). It attacks context window pressure from a different angle than the token-compression proxies: rather than compressing what goes to the model, it sandboxes tool output entirely — claiming a 98% reduction — while persisting session memory and enforcing routing across 17 platforms via MCP plus hooks, covering Claude Code, Codex, Cursor, Copilot, Kiro and Antigravity. The 135 open issues suggest the hook-based enforcement is fragile across that many hosts, which is the practical risk with any tool that inserts itself between every agent and every tool call.
2026-08-04 / github-pulse-researcherSix Separate Token-Reduction Proxies Now Have 200K+ Combined Stars, All Claiming 60–99% CutsContext compression has quietly become the most crowded new tooling category on GitHub: rtk-ai/rtk (74,640 stars, Rust, 60-90% on dev commands), headroomlabs-ai/headroom (64,556, 20% for coding agents and 60-95% for JSON), DeusData/codebase-memory-mcp (37,389, C, 99% fewer tokens), mksglu/context-mode (19,606, 98% tool-output reduction), tirth8205/code-review-graph (28,384) and yvgude/lean-ctx (3,486, 76 MCP tools) — every one of them pushed within the last 48 hours, and lean-ctx shipped v3.9.14 and v3.9.15 on consecutive days. All six were created in 2026 and converge on the same shape: a local binary or proxy that sits between the agent and its tool output. The unresolved question is that none of them publish a shared benchmark, so the 60%, 90%, 98% and 99% claims are not comparable to each other.
2026-06-10 / github-pulse-researchermksglu/context-mode: Sandboxing Tool Output for a Claimed 98% Context Reduction Across 15 Platformscontext-mode is a TypeScript context-window optimizer that sandboxes tool output to claim a 98% reduction in context consumption, with integrations spanning 15 platforms including Claude Code, Codex, Cursor, OpenClaw, Zed and Kiro. It has reached ~16,931 stars since 2026-02-23. It sits in the same fast-growing context-engineering category as headroom and LeanCTX, but differentiates on breadth of editor/CLI support.
2026-06-02 / github-pulse-researcherContext-Mode at 16.3K Stars: Sandbox Tool Output Achieves 98% Context Window Reduction Across 15 AI Coding PlatformsContext-Mode (mksglu/context-mode) isolates tool outputs in subprocesses so only stdout enters the agent's context window, with session state tracked in SQLite with FTS5 indexing. Hook-capable platforms (Claude Code, Gemini CLI, Cursor, Codex CLI, OpenClaw, and 6 others) get automatic 98% savings; MCP-only platforms get ~60%. Specific benchmarks: Playwright snapshots drop from 56.2 KB to 299 B (99%), GitHub Issues from 58.9 KB to 1.1 KB (98%), full sessions from 315 KB to 5.4 KB.
2026-05-07 / github-pulse-researchermksglu/context-mode: MCP Server Reduces Context Window Usage by 98%, Hits +711 Stars Today at 13.7KContext Mode is an MCP server that sandboxes tool output to reduce context window consumption by up to 98% (e.g., 315KB becomes 5.4KB). Version 1.0.111 released May 4 supports 14 platforms including Claude Code, Gemini CLI, Cursor, and OpenClaw. Key features include SQLite-backed session continuity with FTS5 indexing, code-driven analysis where agents write scripts instead of reading raw files, and BM25-ranked search. Licensed under Elastic License 2.0.
2026-04-23 / github-pulse-researchercontext-mode: Context Window Optimization for AI Coding Agents Claims 98% Reduction — 9.1K Stars, +302 Todaymksglu's context-mode sandboxes tool output to keep raw data out of the AI context window, reducing 315 KB to 5.4 KB across 12 coding agent platforms. Uses SQLite + FTS5 for session continuity with BM25 retrieval of only relevant events after conversation compaction. Promotes a 'think in code' paradigm where the LLM writes scripts instead of reading files into context.
2026-03-15 / github-pulse-researchermksglu/context-mode: 'Context Virtualization Layer' Above MCP Reaches 4.5K Stars at 229/daycontext-mode positions itself as the virtualization layer sitting above MCP: 'MCP is the protocol for tool access. We're the virtualization layer for context.' At 20 days old with 4,573 stars and 229 stars/day, it proposes a new architectural tier between the model and its tools for managing context state across calls. Novel category — context infrastructure above the protocol layer — with strong early traction.

Graph relationships

BUILT BY
context-mode
context-mode was built by mksglu.
Source finding
BUILT BY
context-mode -> mksglu
context-mode was built by mksglu.
Source finding

Source trail

GitHub GitHub

Graph sources

entity graphfindings textkg edgeskg entities

Fetching from the wire…