ClawBench is an open-source benchmark that tests whether browser-based AI agents can complete everyday online tasks on live websites. It is used to evaluate agents on activities such as booking travel, ordering food, applying for jobs, and managing email.
Borrowing it
Nothing to install: this file belongs to TIGER-AI-Lab/ClawBench. Take a copy, put it at the same path in your own repository, and replace the rules that are about this project with yours.
curl -O https://raw.githubusercontent.com/TIGER-AI-Lab/ClawBench/main/AGENTS.mdgit clone --depth 1 https://github.com/TIGER-AI-Lab/ClawBenchWrote 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/instructions/tiger-ai-lab/clawbench/agents-md)<a href="https://agentmods.dev/instructions/tiger-ai-lab/clawbench/agents-md"><img src="https://agentmods.dev/badge/instructions/tiger-ai-lab/clawbench/agents-md/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/instructions/tiger-ai-lab/clawbench/agents-md"><img src="https://agentmods.dev/badge/instructions/tiger-ai-lab/clawbench/agents-md.svg" alt="Reviewed on agentmods" width="80" height="20"></a>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 | $0.01826 | $0.01826 |
| Opus 5.5 | $0.00730 | $0.00730 |
| Sonnet 5.5 | $0.00365 | $0.00365 |
| Haiku 4.5 | $0.00183 | $0.00183 |
Grade A, and why
ClawBench AGENTS.md 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 12d 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 — 152 lines — stays where its author put it; the contents beside it link to each section on GitHub.
ClawBench -- Agent Context
This file is for coding agents (Claude Code, Cursor, Copilot, etc.) to understand and operate on the ClawBench project. If you are a human, see the README.
What This Is
ClawBench is a benchmarking framework for evaluating AI web agents on real-world online tasks. V1 lives in test-cases/v1/ with 152 tasks spanning 143 live websites and 15 life categories; V2 lives in test-cases/v2/ with 129 tasks; Lite lives in test-cases/v1-lite/ with 20 link-backed V1 tasks. Each task runs in an isolated Docker container with a browser, a recording Chrome extension, and an AI agent harness (openclaw, opencode, claude-code, claude-code-chrome-extension, codex, browser-use, claw-code, hermes, or pi, or webbrain, selectable via --harness). The framework captures five layers of data: session replay (MP4), action screenshots, HTTP traffic, browser actions, and agent messages.
Project Structure
ClawBench/
run.sh # Entry point -- launches interactive TUI
pyproject.toml # Root uv package metadata and CLI scripts
.env # Shared PurelyMail credentials packaged with the CLI
src/
clawbench/ # Main Python package
tui.py # Interactive TUI (called by run.sh)
runner/
run.py # Single test-case runner
batch.py # Batch runner (model x case cross-product)
run_support/ # Runner-private helpers for single runs
utils/
hf_upload.py # Optional HuggingFace upload helpers
runtime/
shared/
alex_green_personal_info.json
runtime-server/ # Container-only uv project for server deps
chrome-extension/ # Recording extension
harnesses/ # Dockerfiles + setup/run scripts
models/
models.yaml # Model API configs (gitignored -- copy from example)
models.example.yaml # Template with placeholder keys
model.schema.json # JSON schema for model entries
test-cases/
task.schema.json # JSON schema for task.json
v1/ # V1: 152 task directories
v2/ # V2: 129 task directories
v1-lite/ # Lite: 20 V1 task directories with linked task files
eval/
README.md # Evaluation guide + Claude Code prompt template
agentic_eval.md # Evaluator rubric for judging agent success
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.
- 12d ago Changed · +2 lines · +37 tokens per session 5b776df8c305
- 1mo ago Changed · +4 lines · +110 tokens per session bc7e4bd18194
- 2mo ago First seen · 146 lines · 1,679 tokens per session scan A a0a6e404a594
ClawBench AGENTS.md is an instructions file published in the GitHub repository TIGER-AI-Lab/ClawBench (1,147 stars, last pushed 6d ago), licensed Apache-2.0. It adds 1,826 tokens to every session, about $0.0073 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.
Other instructions, from other repositories
vscode buildNext.instructions.md
Working notes and architecture documentation for the new esbuild-based build system in build/next. Use when making changes to the new build pipeline (transpile/bundle commands, NLS plugin, source-map handling, resource copying, or self-hosting watch tasks).
next.js AGENTS.md
AGENTS.md instructions for vercel/next.js, covering next.js development guide, codebase structure, monorepo overview, core package: packages/next and other important packages.
vscode oss-third-party-notices.instructions.md
Instructions for microsoft/vscode, covering vs code oss third-party-notices pipeline, architecture, pipeline flow in ci, cache eligibility and rollout and applying the notice (cutover).
langchain AGENTS.md
AGENTS.md instructions for langchain-ai/langchain, covering global development guidelines for the langchain monorepo, corridor security analysis, project architecture and context, monorepo structure and development tools & commands.
vscode copilot-instructions.md
Copilot instructions for microsoft/vscode, covering vs code copilot instructions, project overview, root folders, core architecture (src/ folder) and built-in extensions (extensions/ folder).
deepseek-harness AGENTS.md
AGENTS.md instructions for deepseek-ai/deepseek-harness, covering agents.md, pre-stable apis and released session data, repository layout, commands and host sandbox failures.