A technical auditing tool for monitoring and investigating token consumption across scheduled cron jobs and heartbeat sessions.
Install & Download
1. ClawHub CLI
The fastest way to install a skill directly from the registry.
npx clawhub@latest install openclaw-scheduler-token-auditor
2. Manual Installation
Copy the skill folder to one of these locations
~/.openclaw/skills/ <project>/skills/ Priority: Workspace > Local > Bundled
3. Prompt Installation
Copy this prompt to OpenClaw to install it automatically.
Help me install openclaw-scheduler-token-auditor using Clawhub. If Clawhub is not installed, install it first (npm i -g clawhub).
Prefer to download?
Get the raw skill files in a ZIP archive.
What is OpenClaw Scheduler Token Auditor?
The openclaw-scheduler-token-auditor is a specialized utility designed for developers and system administrators to maintain strict control over their AI infrastructure costs. It provides a structured methodology to audit Openclaw Skills token usage by differentiating between authoritative measured data, bounded risk assessments, and inconclusive records. This ensures that users can pinpoint exactly which scheduled tasks are inflating their bills and why.
By leveraging the authoritative usage.total_tokens field, this skill prevents the common pitfalls of manual estimation. It offers a transparent look into cron job performance and heartbeat session health, providing clear pass/fail verdicts against user-defined or default token thresholds. This level of granular visibility is essential for maintaining sustainable AI agent operations.
OpenClaw Scheduler Token Auditor Use Cases
- Identifying which specific cron job is consuming the highest volume of tokens.
- Auditing scheduled tasks against a strict budget threshold, such as 50,000 or 100,000 tokens per run.
- Investigating unexpected spikes in scheduler token usage to find the root cause.
- Validating whether heartbeat sessions are operating within expected lightweight bounds.
- Analyzing expensive runs to detect oversized prompts, task scope creep, or inefficient tool loops.
How OpenClaw Scheduler Token Auditor Works
- Target Classification: The skill first categorizes the audit target into either a cron job or a heartbeat session to apply the correct evidence rules.
- Data Retrieval: It fetches recent run records using cron list or session history tools to gather raw usage metrics.
- Evidence Labeling: Every target is assigned an evidence class: Exact (authoritative data exists), Bounded (risk-based assessment), or Inconclusive (missing data).
- Threshold Comparison: The skill compares the total tokens against a defined threshold (defaulting to 50,000) to determine a verdict.
- Cause Analysis: For verified expensive runs, it inspects metadata and transcripts to explain the high consumption, such as repeated file reads or large context injection.
- Standardized Reporting: It outputs a concise report including the target type, evidence class, token finding, and threshold verdict.
OpenClaw Scheduler Token Auditor Setup
To use this Openclaw Skills auditor, ensure your environment is configured with the OpenClaw CLI. You can trigger the audit directly via slash commands or standard natural language queries.
# List available cron jobs to audit
openclaw cron list
# Audit a specific cron job by ID
openclaw cron runs --id <jobId> --limit 50
# Trigger the auditor via natural language
/token_auditor --threshold 80000
OpenClaw Scheduler Token Auditor Data Schema & Taxonomy
The auditor organizes data based on the source of evidence and the certainty of the metrics. It strictly prioritizes the usage.total_tokens field as the single source of truth.
| Evidence Class | Source Data | Reliability | Output Type |
|---|---|---|---|
| Exact | usage.total_tokens | High | Authoritative count and pass/fail verdict |
| Bounded | Session structure/history | Medium | Risk-based statement (e.g., 'rules out 50k+ behavior') |
| Inconclusive | Missing usage fields | Low | Unknown status with structural observations |
OpenClaw Scheduler Token Auditor Advanced Features
- Support for custom token thresholds to align with specific project budgets.
- Detailed heartbeat analysis that rules out extreme usage through structural pattern recognition.
- Identification of specific failure modes like the 'Hand-sum trap' or 'Transcript-as-meter trap'.
- Contextual explanations for expensive runs, identifying issues like tool fan-out or session reuse problems.
- Multi-evidence reporting that prevents the blurring of exact cron data with bounded heartbeat estimates.
SKILL.md
Loading
Related Openclaw Skills
Vyper Reference Tool
A specialized technical reference tool for developers building secure and efficient smart contracts using the Vyper programming language.
SalesBlink
SalesBlink for Openclaw Skills allows AI agents to automate outbound sales, email outreach, and lead generation workflows via the Membrane CLI.
Regex Toolkit
A powerful CLI utility for testing, matching, extracting, and explaining regular expressions with a library of 25+ pre-defined patterns.
SmartReach
A powerful sales engagement skill that allows AI agents to automate multi-channel outreach and manage prospect data through SmartReach.
Mouse & Keyboard Control
A macOS automation skill that enables AI agents to control the cursor and keyboard via cliclick commands.
macOS AppleScript Fallback
A robust automation skill providing reliable shell and AppleScript fallbacks for core macOS productivity applications.
Featured*
