bytedance/agentkit-samples is a collection of examples and tutorials for Volcengine AgentKit, an AI-agent development platform for building, deploying, and operating agent applications. Developers use the samples to learn agent creation, multi-agent collaboration, memory, retrieval, MCP integrations, media generation, customer service, and other workflows. The catalogue skills provide agent workflows based on these examples.
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 bytedance/agentkit-samples --skill byted-ark-agent-eval-labgit clone --depth 1 https://github.com/bytedance/agentkit-samplesWrote 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/bytedance/agentkit-samples/byted-ark-agent-eval-lab)<a href="https://agentmods.dev/skills/bytedance/agentkit-samples/byted-ark-agent-eval-lab"><img src="https://agentmods.dev/badge/skills/bytedance/agentkit-samples/byted-ark-agent-eval-lab/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/bytedance/agentkit-samples/byted-ark-agent-eval-lab"><img src="https://agentmods.dev/badge/skills/bytedance/agentkit-samples/byted-ark-agent-eval-lab.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 | $0.00162 | $0.16194 |
| Opus 5.5 | $0.00065 | $0.06478 |
| Sonnet 5.5 | $0.00032 | $0.03239 |
| Haiku 4.5 | $0.00016 | $0.01619 |
Grade A, and why
byted-ark-agent-eval-lab 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 6d 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 — 851 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Agent Eval Lab 评测与已有结果评分
帮助用户明确评测问题、准备方案、执行评测和解释结果；平台操作只使用 agent-eval CLI。
本文件是操作指南，不是工作流引擎或持久化意图协议。纯方法讨论不要求安装、鉴权或选择项目；
需要真实资源、历史结果或平台写操作时，再检查对应前置条件。
当前 edition 的身份与环境
安装、身份与环境
当前外场 CLI 使用永久 Volcengine AK/SK 对 OpenTOP 请求做 V4 签名。AK/SK 必须由宿主或受控密钥管理 注入执行 CLI 的进程环境：
VOLCENGINE_ACCESS_KEYVOLCENGINE_SECRET_KEY
两项必须同时存在。不得把它们写入 Profile、命令参数、日志、文档或对话，也不得要求用户粘贴明文。 当前 external CLI 不提供交互式登录或登出子命令；不要臆造或尝试调用它们。
如果 agent-eval 尚未安装，先取得外场发行说明中已确认的精确包名、registry 和版本。外场 npm 包名与
公开 registry 尚未冻结时，停止并向发布负责人确认；不得套用内场包名、内网 registry 或移动的
latest/beta 标签。
在托管 Agent 会话中，先检查宿主是否已注入完整凭证，再验证服务链路：
agent-eval --version
agent-eval --help
agent-eval auth status --format json
agent-eval status --format json
auth status 只报告凭证是否成对注入及认证类型，不显示 AK/SK，也不访问服务。未配置时会返回
authenticated=false；只配置一项时命令拒绝并要求宿主修正注入。status 才会使用 V4 签名调用
AgentEvalLabV1GetStatus。凭证缺失、签名失败或服务拒绝时停止并报告稳定错误码和 Request ID，
不得回退内场登录态或旧服务入口。
只有在开发者控制的本地 Shell 中，才可保存非敏感的 Profile 作用域：
agent-eval config init --profile <profile>
agent-eval auth status --format json
agent-eval status --format json
agent-eval projects list --format json
agent-eval projects use <project_id> --format json
agent-eval projects current --format json
Profile 只保存默认 Project，绝不保存 AK/SK。托管 Agent 会话不得执行 config init、
config use、config show、config list 或 projects use；改用显式 --project-id。显式
--project-id 始终优先，不得自动选择项目列表第一项。
OAuth + PKCE 是后续可选能力。在包含该能力的发行版明确提供对应 auth 子命令和验收说明之前，
不得臆造登录、刷新或持久化凭证流程。
当前不接入未被平台支持的自定义／私有 Agent、任意 HTTP 服务或本地仓库作为新 Target。
这不排除用户账号下已有的 MA，也不减少其他已支持 Target 的配置能力；实际类型以
targets agent-types 和 targets spec 为准。用户已有外部输出时可以独立评分，
但不得把它当作自定义 Agent 执行接入的替代实现。
按用户意图进入
| 用户当前需要 | 从哪里开始 / 停在哪里 |
|---|---|
| 讨论评测方法、设计方案 | 先给任务、数据与评分建议；不自动创建、发布或运行 |
| 首次接入 | 安装与诊断 → 选择有权限的 Project |
| 评测涉及业务 CLI／Skill、私有产物或业务登录 | 读取执行准备参考，在发布前主动核对登录需求；明确无需登录不追问，未知时询问一次，再按需绑定项目资产 |
| 保存／修改草稿，先不发布 | 发现所需资产 → 保存 Draft → 回读 revision/digest 和缺项，结束；不强制补齐后发布 |
| 创建并发布方案，先不运行 | Draft → Preview / Validate → Publish → 回读 Snapshot，结束 |
| 运行已发布方案 | 回读精确 Snapshot → 核对授权 → 提交一次 → 保存 ID → 限时观察 |
| 从现有 Snapshot 先试跑少量题目 | 进入“先验证少量题目”，按真实 Snapshot item key 启动选题 Run；不复制数据、不重发方案、不自动全量运行 |
| 新建评测来比较 Agent、模型、Skill 或 Tool | 先区分完整 Target 对比和单变量对比，再进入方案流程；不把整体差异解释成组件因果 |
| 比较已有历史 Run、解释差异、保存或找回实验结论 | 读取历史分析参考；新分析先核对冻结条件、实际样本与评分，获授权后保存并回读；找回直接读取已有归档，不创建新 Comparison 或评测 |
| 只有业务问题／需要导入数据 | 读取数据与评分参考，选择文本草稿或规范包导入；数据发布不等于评测执行 |
| 导入外部 Evidence／给已有输出或轨迹评分 | 进入“导入外部 Evidence 并评分”；已有 Evidence 从评分开始，只导入则不评分 |
| 已有结果，按原规则再评、换评分规则或查看差异 | 读取数据与评分参考，区分独立 Assessment 与原生 Run 重评分；原规则重评优先复用兼容的已发布评分版本，只查差异不创建新评分 |
| 拉全历史重评分、判断评分器稳定性 | 读取历史重评分分析；按来源取全、按证据和评分条件分组，区分同版本重复一致性与跨版本差异，不自动补跑 |
| 查历史结果、解释失败 | 从 Run ID 或 Comparison execution ID 只读查询；需要定位异常或提出修正时读取诊断与修正参考，不自动修改、重建方案或重跑 |
| 已发布 Snapshot 被拒绝，或询问旧版本是否仍可运行 | 读取Snapshot 兼容性诊断；尚无 Run 也可从失败请求取证，不因版本旧就刷新或重跑 |
| 反馈评测问题、连续失败或表达不满 | 读取用户反馈参考；明确要求上报时整理提交，其他情况只询问一次，不静默上传 |
| 修改、取消、只重试部分失败项 | 进入“修改、取消或重新执行”，先区分新版本、取消意图和派生 Run |
| 创建、修改或停用定时评测 | 读取高级操作参考的 Schedule 章节，核对运行时准入（当前不支持 MA 定时执行）、单次／周期授权与精确 Snapshot；停用用 pause，取消某轮 Run 不会停用 Schedule |
What ships with it
8 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.
- 6d ago First seen · 851 lines · 162 tokens per session scan A eaaa2cc13152
byted-ark-agent-eval-lab is a skill published in the GitHub repository bytedance/agentkit-samples (484 stars, last pushed 3d ago), licensed Apache-2.0. It adds 162 tokens to every session and 16,194 once invoked, about $0.0006 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 2026-11-03.
byted-ark-agent-eval-lab compared with skills from other repositories
| Mod | Type | Stars | Last push | Licence | Scan | Tokens | Works with |
|---|---|---|---|---|---|---|---|
| byted-ark-agent-eval-labbytedance/agentkit-samples | Skill | 484 | 3d ago | Apache-2.0 | A | 162 | Codex |
| ensembl-databasesynthetic-sciences/openscience | Skill | 4.4k | yesterday | Apache-2.0 | A | 45 | Claude Code, Codex |
| skill-news-intelligenceZJU-REAL/Easel | Skill | 4.5k | 7d ago | Apache-2.0 | A | 104 | Claude Code, Codex |
| ai-musicZJU-REAL/Easel | Skill | 4.5k | 7d ago | Apache-2.0 | A | 161 | Claude Code, Codex |
| sidewinder-rattlesnakePurpleAILAB/Decepticon | Skill | 6.6k | yesterday | Apache-2.0 | A | 48 | Claude Code |
| 10-batchai-driven-dev/framework | Skill | 552 | yesterday | MIT | A | 47 | Claude Code |
| beevibe-team-mesh-negotiationbeevibe-ai/beevibe | Skill | 341 | 2d ago | Apache-2.0 | A | 112 | Claude Code, Codex |