quota-guard: a free Claude Code plugin that caps re-sent tool output

The replay on the cost-anatomy page says −54% of re-sent context would disappear if tool results were capped at 8,000 characters, and the whole saving sits in 34 heavy sessions holding 50% of billed context. This plugin is that cap, as three hooks. MIT, beta, 19 KB.

PreToolUse · Bashpipes each command's output through a pager that keeps the head and tail and saves the full stream to a file the agent can grep

PreToolUse · Readadds a one-line note when the same unchanged file is read again in a session

/report skillruns the cost anatomy on your own logs and shows what the cap saved

Install (60 seconds)

curl -L https://agentexchange.work/data/quota-guard.zip -o quota-guard.zip
unzip -q quota-guard.zip
claude --plugin-dir ./quota-guard

Cap size: QUOTA_GUARD_CAP=12000 claude --plugin-dir ./quota-guard. Commands that change directory, background commands and already-wrapped commands are passed through untouched. Runs in your own shell (zsh or bash); nothing is sent anywhere. Tests: python3 quota-guard/tests/test_hooks.py.

Download quota-guard.zip (free)

What it does not do

It cannot rewrite MCP tool results (hooks cannot modify results after the fact), it does not change compaction, and it does not touch your account or usage data. The measured ceiling of this approach is on the cost-anatomy page; if your sessions are the light kind, expect a few percent, not half.

The kit: measure your own sessions and the market, weekly

Session cost anatomy for your own logs (quota_report.py): the exact report on this site, on your machine, one command; reads only usage fields and lengths.
Demand simulator (demand_sim.py + priors + calibration): live connectors for GitHub, npm, VS Code Marketplace, WordPress.org, Hacker News, Apify, MCP registry, Hugging Face; scores what a solo newcomer reached at each attach-point in the last 24 months, calibrated against six real product outcomes before it scores anything.
This week's datasets: every table on this site as JSON and TSV, the plugin marketplace catalog, the issue gap maps, the 50-candidate scoring run with every verdict and its deciding number.

$39, one time. Regenerated weekly for a year; the download link you get keeps serving the newest build. MIT.

Get the kit, $39

Card checkout by Stripe on this site's own account; the download page opens the moment payment completes. Questions: riley@agentexchange.work.

Measured, not estimated: each figure names its source and pull date; where a source was unreachable the figure is omitted. Free JSON for every dataset at /api/index.json; agents: MCP endpoint https://agentexchange.work/data/mcp (streamable HTTP, no auth). RSS. Built and run by one person; the $39 kit pays for it.