Tokenoptimizer
PassAudited by ClawScan on May 1, 2026.
Overview
Tokenoptimizer appears to be a disclosed cost-optimization utility, but it can persistently change your OpenClaw configuration and agent prompt/context files when you run it with apply mode.
Use this if you want to change OpenClaw cost settings. Start with `python cli.py optimize` to preview, inspect any generated config and prompt/template files, adjust the budget caps to your own limits, and only then run commands with `--apply`.
Findings (4)
Artifact-based informational review of SKILL.md, metadata, install specs, static scan signals, and capability signals. ClawScan does not execute the skill or run runtime probes.
If you run apply mode, the skill can change your model routing, heartbeat, caching, rate limits, and budget settings, which can affect agent behavior and spending.
The CLI has authority to write the main OpenClaw configuration when not in dry-run mode.
self.config_path = self.openclaw_dir / 'openclaw.json' ... with open(self.config_path, 'w') as f: json.dump(config, f, indent=2)
Run the dry-run preview first, review the diff, and only use `--apply` if the proposed model, heartbeat, cache, and budget settings match your goals.
Persistent prompt/context changes may alter how your agent chooses models, loads context, and follows operating rules in later tasks.
The tool creates or modifies persistent workspace and prompt files that can influence future agent sessions.
`~/.openclaw/workspace/SOUL.md` Agent principles and operating rules ... `~/.openclaw/prompts/OPTIMIZATION-RULES.md` Copy these rules into your agent prompt.
Inspect generated SOUL, USER, and optimization-rules files before relying on them, and use rollback if the changes are not desired.
Users may need to run the included Python CLI manually and should be sure they are using the intended local artifact rather than an untrusted copy.
The registry does not declare an automated install mechanism or source provenance even though the package contains runnable Python code.
Source: unknown ... Install specifications: No install spec — this is an instruction-only skill.
Install from the listed ClawHub/GitHub source you trust, avoid running modified copies, and review the local files before executing the CLI.
A user could assume the tool guarantees a lower bill than the configured budget actually enforces.
The marketing claim emphasizes very large savings, while the generated example budget cap is $200/month, so users should treat savings as estimates rather than guarantees.
Slash your AI costs from $1,500+/month to under $50/month ... "budgets": { "daily": 5.00, "monthly": 200.00 }Set your own daily and monthly budget values after previewing the generated configuration.