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 instructions/aws-bench/aws-bench-datasets/agents-mdgit clone --depth 1 https://github.com/aws-bench/aws-bench-datasetsWrote 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/aws-bench/aws-bench-datasets/agents-md)<a href="https://agentmods.dev/instructions/aws-bench/aws-bench-datasets/agents-md"><img src="https://agentmods.dev/badge/instructions/aws-bench/aws-bench-datasets/agents-md.svg" alt="Measured on agentmods" 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 | $0.01228 | $0.01228 |
| Opus 5 | $0.00614 | $0.00614 |
| Sonnet 5 | $0.00246 | $0.00246 |
| Haiku 4.5 | $0.00123 | $0.00123 |
Grade A, and why
aws-bench-datasets 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 3d 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 — 80 lines — stays where its author put it; the contents beside it link to each section on GitHub.
aws-bench-datasets contributor and agent guide
Purpose
aws-bench-datasets contains the benchmark content executed by the aws-bench framework: task prompts, verifiers, reference solutions, per-task lifecycle hooks, and containerized AWS scenarios. Framework orchestration belongs in the aws-bench repository.
This guide defines the technical contracts published tasks and scenarios must satisfy. If you're proposing or building a new one, start with CONTRIBUTING.md for the contribution process, then come back here for the contracts.
Repository layout
tasks/<scenario-id>/<task-name>/— published task definitions.scenarios/<scenario-id>/— published scenario manifests (scenario.toml), the deployment container and CDK application (scenario/), and lifecycle scripts:deploy/(required), plus optionalreset/,verify/, andcleanup/.shared/judge/— canonical introspection-judge files copied into tasks.shared/tasks/— canonical helpers shared by task hook directories.metric/— dataset-level metric implementation.test/— repository-wide schema, consistency, and reentrancy tests.
Task contracts
Every published task must include task.toml, instruction.md, an environment/, and tests/test.sh; a solution/solve.sh reference solution is optional (the oracle agent uses it when present). The task's [scenario].scenario_id must match a directory under scenarios/.
- Introspection tasks use
metadata.request_type = "introspection", run the agent with an agent role that has a read-only policy (set viaagent_role_nameintask.toml), write a textual answer, and usetests/ground_truth.jsonwith the shared judge. - Mutation tasks use
metadata.request_type = "mutation", run the agent with an agent role that has a write-capable policy (set viaagent_role_nameintask.toml), write the documented structured output, and usetests/check.pyto verify live AWS state. - Keep
[concurrency].mode, the selected agent role, the instruction, the verifier, and the actual behavior of every lifecycle phase consistent. A task is read-only only when its pre-invoke, agent, verifier, and post-invoke behavior are all read-only. - Placeholders in instructions, verifier environments, and ground truth must resolve from scenario exports or the flat string map emitted by pre-invoke.
- Pre-invoke must create
/logs/pre_invoke/placeholder.jsonwhen it outputs placeholders. - Verifiers must create the Harbor reward output.
- Post-invoke must be safe to retry and restore task-specific mutations.
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.
- 3d ago First seen · 80 lines · 1,228 tokens per session scan A e0d7361713a2
aws-bench-datasets AGENTS.md is an instructions file published in the GitHub repository aws-bench/aws-bench-datasets (20 stars, last pushed 7d ago), licensed Apache-2.0. It adds 1,228 tokens to every session, about $0.0061 per session on Opus 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
ClawBench AGENTS.md
Instructions for TIGER-AI-Lab/ClawBench, covering clawbench -- agent context, what this is, project structure, setup and 2. configure at least one model.
aws-bench AGENTS.md
Instructions for aws-bench/aws-bench, covering agents.md, setup, commands, project structure and code style.
OpenCUA CLAUDE.md
Instructions for xlang-ai/OpenCUA, covering claude.md, project overview, commands, environment setup and model inference dependencies.
embodied-eval-automation AGENTS.md
Instructions for Yinzhanqing/embodied-eval-automation, covering repository agent guidance and rules.
aws-bench CLAUDE.md
Instructions for aws-bench/aws-bench, a project described as: aws-bench measures how well AI agents and model combinations perform on real AWS work — diagnosing misconfigurations, provisioning infrastructure, and operating live cloud environments.
langchain-aws AGENTS.md
Instructions for langchain-ai/langchain-aws, covering global development guidelines for the langchain monorepo, project architecture and context, langchain-aws structure, key components and langgraph-checkpoint-aws structure.