kai-os/carnice-agent-trance-prompt-bank

---
pretty_name: Carnice Agent Trace Prompt Bank
license: other
language:
- en
task_categories:
- text-generation
- other
tags:
- agent
- tool-use
- browser
- long-horizon
- prompt-bank
- synthetic
size_categories:
- 1K<n<10K
configs:
- config_name: default
default: true
data_files:
- split: train
path: carnice_trace_prompt_bank_v4.jsonl
- config_name: prompts_only
data_files:
- split: train
path: carnice_trace_prompt_bank_v4.prompts_only.jsonl
- config_name: local
data_files:
- split: train
path: carnice_trace_prompt_bank_v4.local.jsonl
- config_name: web
data_files:
- split: train
path: carnice_trace_prompt_bank_v4.web.jsonl
- config_name: long_horizon
data_files:
- split: train
path: carnice_trace_prompt_bank_v4.long_horizon.jsonl
- config_name: fixtures
data_files:
- split: train
path: carnice_trace_prompt_bank_v4.fixtures.jsonl
---

# Carnice Agent Trace Prompt Bank

This repository is a curated prompt bank for collecting agent traces.

It is **not** a trace dataset by itself. It is the input side: prompts that can be run through an agent harness, then logged into traces with tool calls, observations, and final answers.

The goal of this release is practical:
- keep prompts that work well in an agent harness
- remove prompts that assume hidden local state or user-private state
- expand browser and long-horizon tasks enough to support serious trace collection

## What this contains

The main release is `v4`:

- `carnice_trace_prompt_bank_v4.jsonl`
- full bank with routing metadata
- `4,033` prompts
- `carnice_trace_prompt_bank_v4.prompts_only.jsonl`
- plain `prompt_id` + `prompt`
- `carnice_trace_prompt_bank_v4.local.jsonl`
- `1,983` local/non-web prompts
- `carnice_trace_prompt_bank_v4.web.jsonl`
- `2,050` web prompts
- `carnice_trace_prompt_bank_v4.long_horizon.jsonl`
- `483` harder long-horizon web prompts
- `carnice_trace_prompt_bank_v4.fixtures.jsonl`
- `25` fixture rows for prompts that need seeded local files
- `carnice_trace_prompt_bank_v4.manifest.json`
- counts and source breakdown

## Why this exists

Most public prompt banks are not directly usable for agent trace collection. Common failure modes:
- prompts assume a hidden repo or local file tree
- prompts assume a private account, order, reservation, or inbox
- prompts are benchmark-style but not harness-ready
- browser tasks drift into auth/signup/checkout flows

This dataset was built by curating and filtering several public sources into one bank that works better in a real agent harness.

## How it was built

This release combines prompt material from multiple upstream sources and applies filtering plus routing labels.

High-level process:
1. Start from a narrow local-first prompt core for coding, terminal, planning, and seeded fixtures.
2. Remove broad instruction-following and customer-support style prompt families that are not useful for trace collection.
3. Add filtered browser tasks from WebGym.
4. Add a smaller long-horizon slice from AssistantBench.
5. Remove tasks that require hidden user state, account login, signup, or private context.
6. Keep the result in a format that can be directly run through an agent harness.

## Source mix in `v4`

From `carnice_trace_prompt_bank_v4.manifest.json`:

- `local_gist/terminal_only.accepted`: `1,858`
- `local_execute_code_probe`: `20`
- `Danau5tin/terminal-tasks`: `105`
- `microsoft/webgym_tasks`: `1,945`
- `local_gist/browser_only.accepted`: `72`
- `AssistantBench/AssistantBench`: `33`

## Row schema

Main fields in `carnice_trace_prompt_bank_v4.jsonl`:

- `prompt_id`: stable identifier
- `prompt`: prompt text
- `category`: coarse task family
- `mode`: `local`, `web`, or `local_fixture`
- `requires_web`: whether web access is expected
- `requires_fixture`: whether seeded files are required
- `source`: source family
- `long_horizon`: whether the prompt was tagged as a harder long-horizon task

## How to use it

1. Choose a row from `carnice_trace_prompt_bank_v4.jsonl` or `carnice_trace_prompt_bank_v4.prompts_only.jsonl`.
2. If `requires_fixture=true`, seed the matching files from `carnice_trace_prompt_bank_v4.fixtures.jsonl`.
3. Run the prompt through your agent harness.
4. Save the resulting messages, tool calls, tool outputs, and final answer as the trace.

## Important note on licenses

This release is a curated combination of upstream prompt sources. Source and attribution notes are in `SOURCES.md`.

The overall release is marked `license: other` because it mixes multiple upstream sources with different licensing situations, including user-provided local prompt archives. Review the source notes before republishing or using this release in a stricter commercial/compliance setting.

pretty_name: Carnice智能体轨迹提示词库（Carnice Agent Trace Prompt Bank）
license: 其他许可
language:
- 英语
task_categories:
- 文本生成
- 其他
tags:
- 智能体（agent）
- 工具使用（tool-use）
- 浏览器（browser）
- 长时序（long-horizon）
- 提示词库（prompt-bank）
- 合成数据（synthetic）
size_categories:
- 1K<n<10K
configs:
- config_name: default
default: true
data_files:
- split: train
path: carnice_trace_prompt_bank_v4.jsonl
- config_name: prompts_only
data_files:
- split: train
path: carnice_trace_prompt_bank_v4.prompts_only.jsonl
- config_name: local
data_files:
- split: train
path: carnice_trace_prompt_bank_v4.local.jsonl
- config_name: web
data_files:
- split: train
path: carnice_trace_prompt_bank_v4.web.jsonl
- config_name: long_horizon
data_files:
- split: train
path: carnice_trace_prompt_bank_v4.long_horizon.jsonl
- config_name: fixtures
data_files:
- split: train
path: carnice_trace_prompt_bank_v4.fixtures.jsonl

# Carnice智能体轨迹提示词库（Carnice Agent Trace Prompt Bank）

本仓库为用于收集智能体轨迹的精选提示词库。**其本身并非轨迹数据集**，而是作为输入侧资源：可通过智能体框架运行的提示词，后续可将运行过程中产生的工具调用、观测结果与最终答案一并记录为完整轨迹。

本次发布的核心目标兼具实用性：
- 保留可在智能体框架中稳定运行的优质提示词
- 剔除依赖隐式本地状态或用户私有状态的提示词
- 扩充浏览器任务与长时序任务的覆盖范围，以支撑大规模专业轨迹采集工作

## 数据集内容说明
本次发布的主版本为`v4`：
- `carnice_trace_prompt_bank_v4.jsonl`：包含路由元数据的完整提示词库，共计`4033`条提示词
- `carnice_trace_prompt_bank_v4.prompts_only.jsonl`：仅包含`prompt_id`与`prompt`字段的极简版本
- `carnice_trace_prompt_bank_v4.local.jsonl`：共`1983`条本地/非网页类提示词
- `carnice_trace_prompt_bank_v4.web.jsonl`：共`2050`条网页类提示词
- `carnice_trace_prompt_bank_v4.long_horizon.jsonl`：共`483`条高难度长时序网页类提示词
- `carnice_trace_prompt_bank_v4.fixtures.jsonl`：共`25`条用于依赖预置本地文件的提示词的固定样本
- `carnice_trace_prompt_bank_v4.manifest.json`：包含各类统计数据与来源细分信息的清单文件

## 数据集开发背景
当前多数公开提示词库无法直接用于智能体轨迹采集，常见缺陷包括：
- 提示词依赖隐式仓库或本地文件目录
- 提示词假设用户已拥有私有账户、订单、预约或收件箱
- 提示词为基准测试风格，但无法直接适配智能体框架运行
- 网页类任务涉及登录、注册、结账等非通用流程

本数据集通过筛选并整合多来源公开资源，构建了更适配真实智能体框架运行的提示词库。

## 数据集构建流程
本次发布整合了多个上游来源的提示词素材，并经过筛选与路由标签标注流程，具体步骤如下：
1. 以面向本地场景的窄域核心提示词为基础，覆盖编码、终端操作、规划任务与预置固定样本
2. 移除对轨迹采集无实用价值的通用指令遵循与客服风格提示词族
3. 从WebGym中筛选适配的网页类任务
4. 从AssistantBench中引入少量长时序任务子集
5. 剔除需要隐式用户状态、账户登录、注册或私有上下文的任务
6. 最终输出可直接在智能体框架中运行的标准格式数据

## v4版本的来源构成
根据`carnice_trace_prompt_bank_v4.manifest.json`统计：
- `local_gist/terminal_only.accepted`：`1858`条
- `local_execute_code_probe`：`20`条
- `Danau5tin/terminal-tasks`：`105`条
- `microsoft/webgym_tasks`：`1945`条
- `local_gist/browser_only.accepted`：`72`条
- `AssistantBench/AssistantBench`：`33`条

## 数据行字段规范
`carnice_trace_prompt_bank_v4.jsonl`中的主要字段包括：
- `prompt_id`：稳定唯一标识符
- `prompt`：提示词文本
- `category`：粗粒度任务类别
- `mode`：取值为`local`、`web`或`local_fixture`
- `requires_web`：是否需要网页访问权限
- `requires_fixture`：是否需要预置本地文件
- `source`：来源家族
- `long_horizon`：是否被标记为高难度长时序任务

## 使用指南
1. 从`carnice_trace_prompt_bank_v4.jsonl`或`carnice_trace_prompt_bank_v4.prompts_only.jsonl`中选取目标数据行
2. 若`requires_fixture=true`，则从`carnice_trace_prompt_bank_v4.fixtures.jsonl`中加载对应的预置文件
3. 通过智能体框架运行该提示词
4. 将运行过程中产生的消息、工具调用、工具输出与最终答案保存为完整轨迹

## 许可相关重要说明
本次发布为多上游来源提示词的整合精选集，来源与署名信息详见`SOURCES.md`。
本数据集整体标注为`license: other`（其他许可），原因是其整合了多个不同许可协议的上游资源，包括用户提供的本地提示词归档。在商业发布或严格合规场景下使用前，请务必查阅对应来源的许可说明。