Context windows are being cut by sandboxing the tools. context-mode, out september 7, claims a 98%…
context windows are being cut by sandboxing the tools. context-mode, out september 7, claims a 98% reduction in tool output size by isolating command output, with session memory and routing across 17 platforms via mcp.
the agent's memory diet is the next optimization.
Context
The mksglu/context-mode repository describes context window optimization for AI coding agents that sandboxes tool output (98 percent reduction), persists session memory and enforces routing across 17 platforms via MCP and hooks. It was created on 23 February 2026, its README license badge reads Elastic License 2.0 and GitHub shows Other. The README example turns 315 KB into 5.4 KB, tracks session events in SQLite retrieved with FTS5 and BM25 after compaction, lists 11 MCP tools, and says hook-capable platforms get automatic routing enforcement while other platforms need a one-time routing file copy. It showed 20,331 stars on 4 October 2026.
The note's claims match the maker's own text. The 98 percent is one worked example of tool output and not a measured average across tasks or total agent token use, and no independent benchmark was read. Seventeen platforms is the maker's count and routing is automatic only on hook-capable ones. Out September 7 was not found: the repository was created in February 2026 and the maker blog is dated 26 February. The license is source-available and not OSI open source, and the note does not claim open source. That the memory diet is the next optimization is the author's opinion.
Related work
- Earlier note on context-mode's 98 percent claim ↗Same repository and claim.
- Earlier note on silencing the tools ↗Same repository.
Watch next
- An independent measurement of net token and quality effects across tasks.
Sources
- mksglu/context-mode (GitHub)github.com
Provenance
The note above is reproduced unedited from the original post, first published on Threads on 20 September 2026 at 19:55 IST. Sources are the papers and datasets the note draws on.
View the original post ↗Embed this note
More notes
The air is now being asked to keep its own ledger
the air is now being asked to keep its own ledger: ecmwf’s aifs compo becomes the first ai model to forecast atmospheric composition globally every three hours, cleanair simulates 365 days of pm2.5 over china in ten seconds, and a unified framework maps six pollutants at one kilometer across the whole country. the air now files its own composition report.
read the note →The current is now being asked to draw its own map
the current is now being asked to draw its own map: china’s langya 2.0 predicts six ocean phenomena including internal waves and mesoscale eddies, a deep net called wenhai resolves eddies globally with air sea flux formulas built in, and scripps infers surface currents from the way temperature patterns deform in satellite images. the ocean now files its own circulation report.
read the note →The soil is now being asked to report its own carbon
the soil is now being asked to report its own carbon: a nix color sensor paired with generative data augmentation predicts soil organic carbon without a lab, random forest drives 74 percent of soil health mapping studies, and sentinel 2 tracks five year carbon change across france and italy from 922 samples. the dirt now files its own carbon account.
read the note →