Agents' Last Exam is an evaluation framework and benchmark for testing AI agents on long, economically relevant tasks with outcomes that can be checked. Researchers and developers use its sandboxed tasks, agent runner, and deterministic graders to measure performance across professional work in 55 subdomains and 13 industry groups.
Borrowing it
Nothing to install: this file belongs to rdi-berkeley/agents-last-exam. 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/rdi-berkeley/agents-last-exam/main/AGENTS.mdgit clone --depth 1 https://github.com/rdi-berkeley/agents-last-examWrote 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/rdi-berkeley/agents-last-exam/agents-md)<a href="https://agentmods.dev/instructions/rdi-berkeley/agents-last-exam/agents-md"><img src="https://agentmods.dev/badge/instructions/rdi-berkeley/agents-last-exam/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/rdi-berkeley/agents-last-exam/agents-md"><img src="https://agentmods.dev/badge/instructions/rdi-berkeley/agents-last-exam/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 | $0.01547 | $0.01547 |
| Opus 5.5 | $0.00619 | $0.00619 |
| Sonnet 5.5 | $0.00309 | $0.00309 |
| Haiku 4.5 | $0.00155 | $0.00155 |
Grade A, and why
agents-last-exam 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 10d 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 — 129 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Development Rules
Repository layout
agents-last-exam/
├── ale_run/ Framework code (python -m ale_run is the entry point)
│ ├── agents/ Agent deployers: claude_code, ale_claw, …
│ ├── base_interface/ The contracts: Provider / Executor / Deployer / Trajectory
│ ├── environments/ Providers (gcloud, static) + image registry
│ ├── executors/ Where a deployer runs: sandbox / local / docker
│ ├── orchestration/ Run lifecycle, config loader, factories
│ └── tasks/ Task discovery + driver
├── tasks/ Task packages, grouped by domain (demo/ has the templates)
├── configs/ Reusable agent + environment configs (referenced by path)
├── selected_tasks/ Curated task lists (cli, full, cpu)
├── secret/ .env + GCP key + per-judge eval keys (real values gitignored)
├── docs/ Setup/task/extension guides + the docs/ale-docs-site/ HTML site
├── example_exp.yaml The minimal experiment; start here
└── pyproject.toml uv workspace; Python ≥3.12, <3.14
Conversational Style
- Keep answers short and concise
- No emojis in commits, issues, PR comments, or code
- No fluff or cheerful filler text
- Technical prose only, be kind but direct (e.g., "Thanks @user" not "Thanks so much @user!")
- When the user asks a question, answer it first before making edits or running implementation commands.
- In docs (README,
docs/, user-facing markdown), avoid em dashes (—) when possible; prefer periods, commas, colons, or parentheses instead.
Code Quality
- Read files in full before making wide-ranging changes, before editing files you have not already fully inspected, and when the user asks you to investigate or audit something. Do not rely only on search snippets for broad changes.
- Single-line helper functions with a single call site are forbidden; inline them instead.
- Always ask before removing functionality or code that appears to be intentional.
- Do not preserve backward compatibility unless the user explicitly asks for it.
- Check installed package source (
.venv/lib/python*/site-packages/<pkg>or the editablesubmodules/cua/...paths inpyproject.toml) for external API shapes instead of guessing. - NEVER remove or downgrade code to fix errors from outdated dependencies; upgrade the dependency instead.
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.
- 10d ago Changed · -1 tokens per session a396c80509d1
- 2mo ago First seen · 129 lines · 1,548 tokens per session scan A 232c35c0691d
agents-last-exam AGENTS.md is an instructions file published in the GitHub repository rdi-berkeley/agents-last-exam (1,144 stars, last pushed 2d ago), licensed Apache-2.0. It adds 1,547 tokens to every session, about $0.0062 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.