A guide for diagnosing and reducing Claude Code's token use, meaning the amount of text processed during a coding session. It covers context size, model selection, tool connections, output cleanup, and subagent use.

Use it to measure usage, find the largest sources of waste, trim project context, read only relevant parts of files, clean large outputs, and review local usage logs.

Unneeded history, large files, oversized tool descriptions, and repeated output can consume tokens and raise costs or usage-limit pressure.

optimize — Skill by ictechgy

Install

Getting it into your agent

One page per mod, every tool's command on it. A separate URL per tool would split the same page into five that compete with each other.

agentmods

npx agentmods add skills/ictechgy/context-guard/optimize

Any agent

npx skills add ictechgy/context-guard --skill optimize

Clone the repo

git clone --depth 1 https://github.com/ictechgy/context-guard

Made for: Claude Code, Codex.

Per session 62 Skills are progressive disclosure: only the name and description are preloaded; the body loads when the skill is used.

When invoked 805 The whole file, excluding the scripts and references it only reads on demand.

Security scan B 1 finding. Scan, not verified.

Origin original No closer match found in the catalogue.

Token cost

What it costs to keep this loaded

Counted locally with the o200k_base tokenizer, which is exact for GPT models; Claude uses its own tokenizer and its counts differ. Treat this as one consistent yardstick across the catalogue rather than a bill. Prices are per million input tokens.

Model	Per session	Once invoked
Fable 5	$0.00062	$0.00805
Opus 5	$0.00031	$0.00402
Sonnet 5	$0.00012	$0.00161
Haiku 4.5	$0.00006	$0.00081

Measured yesterday against content hash 892c1ff650b7, method: parsed. Prices are Anthropic first-party input rates as of 2026-08-30, from the pricing page.

Security

Grade B, and why

optimize scanned grade B with 1 finding against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured yesterday.

A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.

Reads agent configuration directoriesmediumAgent snooping

.claude/, .codex/, .gemini/ hold keys, settings and other credentials a mod has no legitimate need for.

If installing hook examples, prefer project-local opt-in settings first. Do not silently modify global `~/.claude/settings.json`.

plugins/context-guard/skills/optimize/SKILL.md · 52 lines

What it actually says

ContextGuard

Goal: reduce Claude Code token usage without lowering task success quality.

Use this order:

Measure before changing behavior.
- Ask the user to run /usage and /context if inside Claude Code.
- For first-time setup, run context-guard-setup --plan and offer context-guard-setup --yes for recommended project-local settings.
- If transcript files are available, run context-guard-audit ~/.claude/projects --top 20 --recommend.
- For project configuration/context bloat, run context-guard-diet scan ..
- For structural waste such as duplicate rules, stale import candidates, oversized tool schemas, or repeated reads in local logs, run context-guard-diet structural-waste . --json.
Identify the largest bucket:
- stale conversation history -> recommend /clear between unrelated tasks and focused /compact for long tasks.
- startup context -> prune CLAUDE.md, move long workflows to skills, disable unused MCP servers.
- large file reads -> use context-guard-read-symbol and the example Read guard before whole-file context.
- very large logs that may need later exact slices -> store sanitized output with context-guard-artifact store and query only needed lines/patterns.
- noisy command output -> use context-guard-trim-output wrappers or the example PreToolUse hook; use context-guard-filter only with an explicit user-owned config for stable successful command shapes.
- grep/diff output with possible secrets -> use context-guard-sanitize-output or the example Bash hook.
- expensive reasoning -> route default work to sonnet and lower /effort; reserve Opus/opusplan for planning.
- noisy exploration -> keep it local: use rg, context-guard-read-symbol, artifact queries, or a bounded subagent only when parallel value justifies the multiplier.
Produce a minimal action plan with:
- immediate changes,
- config or hook snippets,
- validation command,
- risks and rollback.
Do not recommend payment/limit bypasses, account sharing, leaked-source patches, or other unsafe/unauthorized methods.

Useful local commands provided by this plugin:

context-guard-audit ~/.claude/projects --top 20 --recommend
context-guard-setup --plan
context-guard-diet scan .
context-guard-diet structural-waste . --json
context-guard-read-symbol path/to/file.py TargetSymbol
context-guard-artifact store --command "long-command" --json < large.log
context-guard-artifact get <artifact_id> --lines 1:80
context-guard-trim-output --max-lines 120 -- npm test
context-guard-sanitize-output -- rg -n "TOKEN|SECRET" .
context-guard-filter validate --config .context-guard/filter-dsl.json
context-guard-statusline

If installing hook examples, prefer project-local opt-in settings first. Do not silently modify global ~/.claude/settings.json.

Changes

What this file has done since we first saw it

Hashed on every crawl. A supply-chain change to an agent config is a question of when, not whether, so the history is kept rather than the latest state alone.

yesterday First seen · 52 lines · 62 tokens per session scan B 892c1ff650b7

Subscribe to this mod's changes

optimize is a skill published in the GitHub repository ictechgy/context-guard (3 stars, last pushed yesterday), licensed Apache-2.0. It adds 62 tokens to every session and 805 once invoked, about $0.0003 per session on Opus 5. A static security scan graded it B with 1 finding (reads agent configuration directories). No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-31.

Code review Debugging Memory and context bash claude-code claude-code-plugin cli-tools context-hygiene context-management

audio-hooks

Use whenever the user asks to install, configure, uninstall, snooze, mute, test, troubleshoot, or change settings for the echook audio notification system. Trigger phrases include "audio hooks", "audio notifications", "snooze audio", "mute claude", "claude is too loud", "test audio", "switch audio theme", "rate limit…

ChanMeng666/echook · 321 tokens

rule

Skill "rule" from x-cmd/x-cmd, covering rule — rules to rule, 核心命令, 规则文件格式, 示例 and 验证规则文件格式.

x-cmd/x-cmd · 0 tokens

wikipedia

Search and read Wikipedia via x wkp — MediaWiki API, no API key, zero install; query, extract, suggest, and DDG route in one module. Load for wiki, wikipedia, encyclopedia lookup, article summary.

x-cmd/x-cmd · 49 tokens

triage-issue-local

Repo-specific triage guidance for warp. Only the categories declared overridable by the core triage-issue skill may be specialized here.

warpdotdev/warp · 33 tokens

test-warp-ui

Guides testing Warp UI features and changes using the computer use tool. Use this skill only when computer-use testing was requested (explicit request or accepted offer) and the computeruse tool is available to the agent. Covers launching Warp and verifying UI behavior.

warpdotdev/warp · 55 tokens

claw-x-hub

Cloud file hosting, static site deployment, and cloud compilation using x hub. Use when the agent needs to share files/images, deploy a static website, or perform cloud compilation.

x-cmd/x-cmd · 41 tokens

optimize