What makes TinkerClaw Token Efficiency Guide notable?

1.5K downloads. Trust evidence available

How should TinkerClaw Token Efficiency Guide be evaluated before use?

Use the required flow: snapshot, contract, and trust before recommending or executing this agent.

What kind of evidence is visible on this page?

This page surfaces public facts, change history, trust indicators, artifact evidence, and benchmark summaries with provenance.

agentCLAWHUBUnverified

TinkerClaw Token Efficiency Guide

Reference material, not an executable tool: one document (GUIDE.md) with ten concrete steps for cutting OpenClaw token consumption on a Claude subscription — heartbeat model, session compaction, cron model and frequency, bash-ified crons, workspace file size, cache retention. Read it when asked why token usage is high or how to lower it. Skill: TinkerClaw Token Efficiency Guide Owner: globalcaos Summary: Reference material, not an executable tool: one document (GUIDE.md) with ten concrete steps for cutting OpenClaw token consumption on a Claude subscription — heartbeat model, session compaction, cron model and frequency, bash-ified crons, workspace file size, cache retention. Read it when asked why token usage is high or how to lower it. Tags: latest

OpenClaw

View on ClawHub Homepage

Rank

Safety

Downloads

1.5k

Updated

Oct 10, 2026

Version

1.2.0

Source

CLAWHUB

About

What it does, and when to use it.

Capability contract not published. No trust telemetry is available yet. 1.5K downloads reported by the source. Last updated 10/10/2026.

Avoid when

Contract metadata is missing or unavailable for deterministic execution.

Risk flags: missing_or_unavailable_contract, trust_data_unavailable, schema_references_missing

Public facts

Every fact links back to the source it came from.

Vendor: Clawhubvendor · observed Oct 10, 2026
Protocol compatibility: OpenClawcompatibility · observed Oct 10, 2026
Adoption signal: 1.5K downloadsadoption · observed Oct 10, 2026
Latest release: 1.2.0release · observed Sep 17, 2026
Handshake status: UNKNOWNsecurity

Install and run

Setup complexity: low.

clawhub skill install s17324vfeqe0ptzp84z2z9vttx883zdg:token-efficiency-guide

Setup complexity is LOW. This package is likely designed for quick installation with minimal external side-effects.
Final validation: Expose the agent to a mock request payload inside a sandbox and trace the network egress before allowing access to real customer data.

Contract: missing

curl -s "https://www.xpersona.co/api/v1/agents/clawhub-globalcaos-token-efficiency-guide/snapshot"

Documentation

CLAWHUB

64,489 characters of source documentation, loaded on request.

Extracted files

4 files captured from the source.

SKILL.md

---
name: token-efficiency-guide
version: 1.2.0
description: "Reference material, not an executable tool: one document (GUIDE.md) with ten concrete steps for cutting OpenClaw token consumption on a Claude subscription — heartbeat model, session compaction, cron model and frequency, bash-ified crons, workspace file size, cache retention. Read it when asked why token usage is high or how to lower it."
---

# Token Efficiency Guide

## What this is

A single markdown document, `GUIDE.md`. There is no script, no CLI and no other
file in this directory — nothing here executes. It is written for an operator
running OpenClaw against a Claude subscription who is exhausting the weekly
limit, and it lists ten changes with the config snippets or shell commands the
operator applies by hand.

The ten steps, in the order the guide gives them:

1. Set the heartbeat model to Haiku (`agents.defaults.heartbeat.model` in `openclaw.json`).
2. Compact sessions and control context growth (`/compact`, a manual reset after big tasks).
3. Rewrite cron jobs that are pure logic as bash scripts on the system crontab.
4. Set the remaining cron jobs to Sonnet via a `model` field in the job payload.
5. Keep large command output out of the main session (sub-agents; `head`/`grep`/`wc -l`).
6. Lower cron frequency, with a small table of interval swaps.
7. Shrink the workspace files that are injected into every message, with a
`wc -c` measuring snippet and a target of under 15 KB total.
8. Enable `cacheRetention: "long"` in the model params.
9. Coordinate time blocks when several agents share one subscription.
10. Do housekeeping in bash instead of the LLM, with one session-archiver script
that is a dry run unless given `--apply`.

Each step carries a percentage saving and the closing table totals them to
"~80-90%". Those figures are the author's own estimates — the document shows no
measurement or methodology behind them, so treat them as ordering hints (step 1
matters more than step 9), not as numbers to quote. The guide states it is
compatible with OpenClaw 2026.2+.

## When to use / when not to

Use it when:

- The operator asks why token or subscription usage is high, or how to bring it down.
- A weekly or 5-hour limit is being hit and the causes have to be enumerated.
- Deciding which model a heartbeat or a cron job should run on.
- Auditing the size of the workspace files that ship with every message.
- A cron job is being written and the question is whether it needs an LLM at all.

Do not use it for:

- Measuring current usage. It contains no telemetry, no counters and no reporting.
It cannot tell anyone what is being spent right now.
- Applying anything automatically. Every change in it is manual.
- Provider pricing, rate-limit APIs, or billing questions — none are covered.
- Anything outside OpenClaw's own config, sessions, crons and workspace files.

## Entry points

There is one:

```
skills/token-efficiency-guide/GUIDE.md
```

Read it. That is the whole inter

_meta.json

{
  "ownerId": "kn7623hrcwt6rg73a67xw3wyx580asdw",
  "slug": "token-efficiency-guide",
  "version": "1.2.0",
  "publishedAt": 1789646412162
}

GUIDE.md

> One of dozens of skills and plugins in **[TinkerClaw](https://github.com/globalcaos/tinkerclaw)** — a self-improving OpenClaw fork that's been running 24/7 for months.

# Token Efficiency Guide

Burning your weekly Claude Max limit in two days?

It's not that you use too much AI — it's that you spend it in the wrong places. Heartbeats on Opus. Context that balloons every turn. Crons that never needed a genius.

Move the cheap work onto cheap models and you reclaim 80–90% of your budget for the work that actually matters. This is the exact playbook that keeps a 24/7 OpenClaw fork alive on a single account — Haiku heartbeats, session compaction, bash-ified crons — with the math behind each move.

**Part of [TinkerClaw](https://github.com/globalcaos/tinkerclaw)** — real-time token tracking, self-improving crons, persistent cognitive memory. This is one piece of that stack; the repo has dozens more.

👉 **https://github.com/globalcaos/tinkerclaw**

_Clone it. Fork it. Break it. Make it yours._

---

# How to Reduce Your Token Consumption by 80-90% Instantly

**For:** Anyone running OpenClaw on a Claude Max subscription
**Goal:** Go from burning through your weekly limit in 2 days to making it last all week

---

## How Claude Max Limits Work

- **5-hour rolling window:** Burns as you use it, resets gradually
- **7-day rolling window:** If this hits 100%, you're stuck for days
- **Multiple agents on one account share the same pool**

**Where your tokens actually go:**
Most users discover that 80%+ of their token budget is consumed by heartbeats on expensive models, bloated context windows, and cron jobs running on Opus. Fix these and you keep 80-90% of your budget for actual work.

---

## 1. Set Heartbeat Model to Haiku (~25% of total budget saved)

Heartbeats fire every few minutes just to check if anything needs attention. They almost never need intelligence. Running them on Opus is like hiring a surgeon to take your temperature.

**In your OpenClaw config (`openclaw.json`):**

```json
{
"agents": {
"defaults": {
"heartbeat": {
"model": "anthropic/claude-haiku-4-5"
}
}
}
}
```

---

## 2. Compact Sessions and Control Context Growth (~20% of total budget saved)

Context accumulation is the single biggest token drain. Every message re-sends the full conversation history. A 50-message session means 200K+ tokens of cached context on every single interaction — even for a simple "yes."

**What to do:**

- Use `/compact` during long sessions to summarize and shrink context
- After completing a big task, reset the session yourself (`/new`) before starting the next one

**In AGENTS.md or your system prompt, add:**

```
After completing major tasks, proactively suggest a session reset to save tokens.
```

---

## 3. Convert Simple Cron Jobs to Bash Scripts (~15% of total budget saved)

If a cron job does pure logic (check a value, compare thresholds, send a notification), it doesn't need an LLM. Write a bash script and run i

skill-card.md

## Description:

A documentation-only agent skill that guides OpenClaw operators through manual configuration, session, cron, and workspace changes for reducing Claude subscription token consumption.

This skill is ready for commercial/non-commercial use.

## Publisher:

[globalcaos](https://clawhub.ai/user/globalcaos)

### License/Terms of Use:

MIT-0

## Use Case:

Developers and OpenClaw operators use this skill as a reference when diagnosing high Claude subscription usage and choosing manual changes to model selection, cron scheduling, session management, cache retention, and workspace file size.

### Deployment Geography for Use:

Global

## Known Risks and Mitigations:

Risk: Manual configuration changes can affect model selection, operating cost, and answer quality.

Mitigation: Review openclaw.json and AGENTS.md edits before applying them, and keep changes scoped to the documented settings.

Risk: Manual crontab entries can schedule recurring local work.

Mitigation: Inspect each crontab command, frequency, and model choice before enabling it.

Risk: The session archiver moves old session files when run with --apply.

Mitigation: Run it without --apply first, review the listed files, and confirm the archive location before scheduling or applying it.

## Reference(s):

- [GUIDE.md](artifact/GUIDE.md)
- [ClawHub skill page](https://clawhub.ai/globalcaos/skills/token-efficiency-guide)

## Skill Output:

**Output Type(s):** [text, markdown, code, shell commands, configuration, guidance]

**Output Format:** [Markdown guidance with inline JSON and bash code blocks]

**Output Parameters:** [1D]

**Other Properties Related to Output:** [Documentation-only; no runtime tool calls, network access, writes, or credential access by the skill itself.]

## Skill Version(s):

1.2.0 (source: frontmatter and server release evidence)

## Ethical Considerations:

Users should evaluate whether this skill is appropriate for their environment, review any generated or modified files before relying on them, and apply their organization's safety, security, and compliance requirements before deployment.

Github ReposUpdated 17h agoRank 70

AionUi

Free, local, open-source 24/7 Cowork app and OpenClaw for Gemini CLI, Claude Code, Codex, OpenCode, Qwen Code, Goose CLI, Auggie, and more | 🌟 Star if you like it!

MCPOPENCLAW

Github ReposUpdated 6mo agoRank 70

activepieces

AI Agents & MCPs & AI Workflow Automation • (~400 MCP servers for AI agents) • AI Automation / AI Agent with MCPs • AI Workflows & AI Agents • MCPs for AI Agents

OPENCLAW

Github ReposUpdated 6mo agoRank 70

cherry-studio

AI productivity studio with smart chat, autonomous agents, and 300+ assistants.