LongHorizon-Harness is a computer-use harness that lets AI agents continue work across desktop applications and the command line for extended periods by planning, acting, verifying, checkpointing, and recovering. It is for users who need Claude Code, Codex, OpenCode, or DeepSeek Harness to make reliable progress on complex long-running workflows without training a new model. The catalogue entries provide skills for operating this execution loop.
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 skills add AMAP-ML/LongHorizon-Harness --skill weavebench-cua-reproducegit clone --depth 1 https://github.com/AMAP-ML/LongHorizon-HarnessWrote this? Show the measurements
A badge with what this costs and how it scanned, read live from this page, so it follows the numbers instead of freezing them. Markdown for a README, HTML for a documentation site or a project page.
[](https://agentmods.dev/skills/amap-ml/longhorizon-harness/weavebench-cua-reproduce)<a href="https://agentmods.dev/skills/amap-ml/longhorizon-harness/weavebench-cua-reproduce"><img src="https://agentmods.dev/badge/skills/amap-ml/longhorizon-harness/weavebench-cua-reproduce/github.svg" alt="Measured on agentmods" height="20"></a>Or the 80×15 button, for a site that already has a row of RSS and ATOM ones. Only the verdict fits; the numbers stay here.
<a href="https://agentmods.dev/skills/amap-ml/longhorizon-harness/weavebench-cua-reproduce"><img src="https://agentmods.dev/badge/skills/amap-ml/longhorizon-harness/weavebench-cua-reproduce.svg" alt="Reviewed on agentmods" width="80" height="20"></a>- NVIDIA SkillSpector pass
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 5.1 | $0.00081 | $0.01568 |
| Opus 5.5 | $0.00032 | $0.00627 |
| Sonnet 5.5 | $0.00016 | $0.00314 |
| Haiku 4.5 | $0.00008 | $0.00157 |
Grade A, and why
weavebench-cua-reproduce 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 2mo 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.
How it starts
The opening of the file, as written. The whole thing — 178 lines — stays where its author put it; the contents beside it link to each section on GitHub.
WeaveBench CUA-Harness Reproduction
Use this skill to guide an agent through a complete, reproducible WeaveBench CUA-Harness run. The expected repository layout is:
WeaveBench-harness/
WeaveBench/
cua-harness/
skills/weavebench-cua-reproduce/
Do not invent alternate launch commands. Prefer the bundled helper script and the project scripts under WeaveBench/scripts/.
Supported Scope
Supported automation:
- Local Docker/KVM WeaveBench evaluation.
- Qwen 3.7-Plus through an Anthropic-compatible endpoint.
- Claude Code execution backend through
cua_harness_claudecode. - OpenClaw judge setup and verification.
- Official WeaveBench assets downloaded from the HuggingFace dataset.
- A 120G copy of the official Ubuntu qcow2 plus per-task rootfs growth.
Not automated:
- Cloud VM provisioning.
- Public image-hosting service deployment.
- New model-provider adapters beyond the existing Anthropic-compatible path.
- Non-Claude-Code execution backend reproduction.
- Deleting or pruning existing results.
If the user requests unsupported automation, explain the boundary and provide the closest supported local Docker/KVM path.
Start With Intake
Before doing setup work on a new machine, run or mentally follow:
./skills/weavebench-cua-reproduce/scripts/reproduce.sh intake
Confirm:
- Linux Docker/KVM is available or installable.
- The user accepts large downloads and creation of
Ubuntu_120G.qcow2. - The model endpoint and API key are available.
- Image proxy should be enabled with public upload/show URLs, or disabled because the endpoint accepts large base64 screenshots.
- The desired scope is
doctor,smoke, subset, or full 114-task evaluation. - Long-running/full evaluation is allowed, especially if other experiments may already be active.
Workflow
- Locate the project root. It must contain both
WeaveBench/andcua-harness/. - Read
references/configuration.mdwhen you need exact defaults, environment variables, or paths. - Read
references/assets.mdbefore downloading or validating WeaveBench tasks, runtime assets, judge templates, or VM files. - Read
references/verify.mdbefore reporting whether setup or a run is actually verified. - Read
references/troubleshooting.mdwhen setup, VM, API, image proxy, warmup, or judge errors occur. - Ask the user for missing API details only when they are required and not discoverable from the environment.
- Run commands from the project root unless a command explicitly changes into
WeaveBench/.
What ships with it
6 files beside SKILL.md in the same directory: the scripts, references and assets a skill reads on demand. Not counted in the per-session cost; read them before you install if any of them is executable.
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.
- 2mo ago First seen · 178 lines · 81 tokens per session scan A 3ad149599e4f
weavebench-cua-reproduce is a skill published in the GitHub repository AMAP-ML/LongHorizon-Harness (1,850 stars, last pushed 16d ago), licensed MIT. It adds 81 tokens to every session and 1,568 once invoked, about $0.0003 per session on Opus 5.5. 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.
weavebench-cua-reproduce compared with skills from other repositories
| Mod | Type | Stars | Last push | Licence | Scan | Tokens | Works with |
|---|---|---|---|---|---|---|---|
| weavebench-cua-reproduceAMAP-ML/LongHorizon-Harness | Skill | 1.9k | 16d ago | MIT | A | 81 | Codex |
| open-code-reviewalibaba/open-code-review | Skill | 50k | yesterday | Apache-2.0 | A | 98 | Cursor |
| harness-creatorwalkinglabs/learn-harness-engineering | Skill | 23k | 15d ago | MIT | A | 142 | Codex |
| dashboard-chat-best-practicesvercel/eve | Skill | 5.8k | yesterday | Apache-2.0 | A | 84 | Codex |
| eve-chat-template-evevercel/eve | Skill | 5.8k | yesterday | Apache-2.0 | A | 75 | Codex |
| penguin-pluginPrism-Shadow/penguin-harness | Skill | 3.2k | yesterday | Apache-2.0 | A | 81 | Claude Code, Codex |
| company-employeePrism-Shadow/penguin-harness | Skill | 3.2k | yesterday | Apache-2.0 | A | 85 | Claude Code, Codex |