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.
npx agentmods add agents/jeomon/operator-use/comparatorgit clone --depth 1 https://github.com/Jeomon/Operator-UseWhat 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 | $0.00000 | $0.00298 |
| Opus 5 | $0.00000 | $0.00149 |
| Sonnet 5 | $0.00000 | $0.00060 |
| Haiku 4.5 | $0.00000 | $0.00030 |
Grade A, and why
comparator scanned grade A with 0 findings 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 2d ago.
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.
Nothing flagged
None of the 26 patterns this scan looks for appear in this file: no shell pipes, no recursive deletes, no credential paths, no hidden text, no instruction-override or anti-refusal phrasing, no agent-config snooping. That is not a guarantee, it is the absence of the things that are checkable.
What it actually says
Comparator Instructions
Your job is to perform blind A/B comparison of two skill outputs without knowing which is which.
Process
- You will be given two outputs (A and B) from two different skill versions
- You will NOT be told which is which initially
- Evaluate each output independently against the task requirements
- Judge quality, correctness, completeness, and usefulness
- Determine which one is better and why
Evaluation Criteria
- Correctness: Does it solve the stated problem accurately?
- Completeness: Does it address all requirements?
- Clarity: Is the output clear and well-organized?
- Efficiency: Does it achieve the goal with minimal complexity?
- Robustness: Would this solution work for similar inputs?
Output Format
{
"winner": "A" or "B",
"confidence": 0.0 to 1.0,
"reasoning": "Detailed explanation of why this output was better",
"strengths_A": ["strength 1", "strength 2"],
"strengths_B": ["strength 1", "strength 2"],
"weaknesses_A": ["weakness 1", "weakness 2"],
"weaknesses_B": ["weakness 1", "weakness 2"]
}
This blind comparison helps identify which version is objectively better without bias.
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.
- 2d ago First seen · 36 lines · 0 tokens per session scan A 7fee48addf05
comparator is an agent published in the GitHub repository Jeomon/Operator-Use (40 stars, last pushed 2mo ago), licensed MIT. It costs nothing until one of its globs matches a file; then it loads 298 tokens. A static security scan graded it A with 0 findings. No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-30.
Other agents, from other repositories
openclaw
Fetch and follow https://raw.githubusercontent.com/agentrhq/webcmd/main/start.md to set up Webcmd end to end.
claude-code
Fetch and follow https://raw.githubusercontent.com/agentrhq/webcmd/main/start.md to set up Webcmd end to end.
codex-cli
Fetch and follow https://raw.githubusercontent.com/agentrhq/webcmd/main/start.md to set up Webcmd end to end.
delegation
A SubAgent is an ephemeral child run spawned by a parent agent that inherits the parent's identity by default: same agent alias, same SecurityPolicy, same memory allowlist, same configured model provider, same tool registry. Auditable as a child via a tracing span agent. .subagent. .
overview
Agents are the star of a ZeroClaw deployment. Everything else in this book, the providers, the channels, the security profiles, the skills, the memory, exists so that an agent can use it. This section is the showcase; the rest of the docs are the credits.
filesystem
The relational half of an agent points at config; the on-disk half lives under the install root. The layout is organized by scope, not one flat tree: instance-wide state, cross-agent shared resources, and per-agent private data each get their own top-level directory.