twg-bench-lite
Compare original and translation side by side
🇺🇸
Original
English🇨🇳
Translation
Chinesetwg-bench-lite
twg-bench-lite
Run one read-only prompt twice: once with free Atlassian/local MCP context and
once with paid Atlassian Teamwork Graph context through TWG CLI. This is a
skill-first workflow with a thin bridge. The benchmark
repo owns reusable runner, judge, and report-template code.
twg benchmark lite run将同一个只读提示词运行两次:一次使用免费Atlassian/本地MCP上下文,另一次通过TWG CLI使用付费Atlassian Teamwork Graph上下文。这是一个以Skill为核心的工作流,通过简洁的命令桥接。基准测试仓库包含可复用的运行器、评判器和报告模板代码。
twg benchmark lite runCLI launcher fallback
CLI启动器回退方案
Run . On shell , use
(macOS/Linux) / (PowerShell), then
tell user to add that directory to PATH. Do not treat auth or command errors as
PATH failures.
twg <command>command not found$HOME/.local/bin/twg$env:LOCALAPPDATA\Programs\twg\bin\twg.exe运行。如果shell提示,请使用(macOS/Linux)或(PowerShell),然后告知用户将该目录添加到PATH中。请勿将认证或命令错误视为PATH配置问题。
twg <command>command not found$HOME/.local/bin/twg$env:LOCALAPPDATA\Programs\twg\bin\twg.exeRuntime Support
运行时支持
V1 supports Codex and Rovo live runs. Claude is V2. Do not require users to
install the benchmark CLI locally.
V1版本支持Codex和Rovo实时运行。Claude支持将在V2版本中提供。无需用户在本地安装基准测试CLI。
Eligibility
适用场景
Use a text-only read task both arms can attempt. Do not compare mutations; for
writes, compare a read-only plan or dry-run only after user approval. Preserve
the underlying prompt byte-for-byte for both arms.
使用双方均可尝试的纯文本只读任务。请勿对比写入操作;若涉及写入,仅在用户批准后对比只读计划或试运行版本。确保两个测试分支使用完全相同的提示词(字节级一致)。
Prerequisite
前置条件
For the default full lifecycle command, verify the portable runner and selected
runtime first:
bash
twg benchmark lite --checkThe default check is Codex; use when the run should use Rovo.
--agent rovoFor Rovo runs:
bash
twg benchmark lite --check --agent rovoThe quality judge defaults to the selected runtime, so Rovo runs use a Rovo
judge by default. If the user explicitly wants a different judge runtime, use:
bash
twg benchmark lite --check --agent rovo --judge-agent codexThe managed custom agents are only required for manual orchestration:
twg-bench-lite-controltwg-bench-lite-test
If a manual run needs either agent and it is missing, tell the user to run:
bash
twg skills install --yesThen start a fresh Codex thread so agent discovery refreshes. Do not substitute
generic agents for manual orchestration because arm isolation is the point.
Before a live run, verify setup with:
bash
twg benchmark lite --checkThis checks the bundled runner artifact and the selected live-run runtime.
Codex checks include the Codex executable and Codex home. Rovo checks include
the Rovo executable and OAuth status. If differs from ,
the check verifies both runtimes. It also reports the global
skill install and managed control/test Codex agent configs as manual-workflow
optional checks when relevant.
--judge-agent--agenttwg-bench-lite对于默认的全生命周期命令,请先验证可移植运行器和所选运行时:
bash
twg benchmark lite --check默认检查针对Codex;若运行需使用Rovo,请添加参数。
--agent rovo针对Rovo运行:
bash
twg benchmark lite --check --agent rovo质量评判器默认使用所选运行时,因此Rovo运行默认使用Rovo评判器。如果用户明确希望使用不同的评判器运行时,请使用:
bash
twg benchmark lite --check --agent rovo --judge-agent codex仅在手动编排时才需要托管自定义Agent:
twg-bench-lite-controltwg-bench-lite-test
如果手动运行需要其中任一Agent但未安装,请告知用户运行:
bash
twg skills install --yes然后启动新的Codex线程以刷新Agent发现。请勿使用通用Agent替代手动编排,因为分支隔离是核心要求。
在实时运行前,请使用以下命令验证设置:
bash
twg benchmark lite --check该命令会检查捆绑的运行器工件和所选的实时运行时。Codex检查包括Codex可执行文件和Codex主目录。Rovo检查包括Rovo可执行文件和OAuth状态。如果与不同,检查将验证两个运行时。此外,当涉及手动工作流时,它还会报告全局 Skill安装情况以及托管的控制/测试Codex Agent配置(作为可选检查项)。
--judge-agent--agenttwg-bench-liteArms
测试分支
- Control - Free Atlassian with Local MCPs Context: use free Atlassian tools plus user-local MCPs/connectors. Do not use TWG CLI, TWG skills, paid graph tools, prior artifacts, or local files as task evidence unless the prompt is explicitly about local files.
- Test - Paid Atlassian Teamwork Graph Context: use TWG CLI graph context with bundled TWG guidance as the primary work-data source.
- 控制分支 - 免费Atlassian搭配本地MCP上下文:使用免费Atlassian工具加上用户本地MCP/连接器。除非提示词明确涉及本地文件,否则请勿使用TWG CLI、TWG Skills、付费图谱工具、先前工件或本地文件作为任务证据。
- 测试分支 - 付费Atlassian Teamwork Graph上下文:使用TWG CLI图谱上下文,并将捆绑的TWG指导作为主要工作数据源。
Paired Run
成对运行
For the default full lifecycle, run:
bash
twg benchmark lite run --prompt "<read-only prompt>" --output-dir <dir>Use to run the same lifecycle with Rovo.
The quality judge uses the same runtime unless is
supplied.
--agent rovo--judge-agent codex|rovoFor manual orchestration:
- Save the exact comparison prompt.
- Ask TWG CLI for the test-arm route before launching the paired agents:
.
twg benchmark lite plan --prompt-file <prompt-file> --json - Spawn both managed agents concurrently with no forked parent context:
and
twg-bench-lite-control.twg-bench-lite-test - Send the identical prompt as the complete user message to each child. Do not add arm-specific instructions; the managed agent configs own isolation. The test agent may receive the TWG route plan as tool-selection policy only; it is not answer evidence.
- Leave model/reasoning overrides unset unless the user asks; apply any override equally to both arms.
- Wait for both children to finish and record their agent IDs or session links.
- Do not solve the task again in the parent. The parent compares, evaluates, and reports.
Print lifecycle progress in the parent session: start, prompt captured, agents
started, each arm finished, outputs read, quality reviewed, report compiled,
done.
对于默认的全生命周期运行,请执行:
bash
twg benchmark lite run --prompt "<read-only prompt>" --output-dir <dir>添加参数可使用Rovo执行相同的生命周期流程。质量评判器默认使用与运行时相同的环境,除非指定参数。
--agent rovo--judge-agent codex|rovo对于手动编排:
- 保存精确的对比提示词。
- 在启动成对Agent前,向TWG CLI请求测试分支的路由:。
twg benchmark lite plan --prompt-file <prompt-file> --json - 同时启动两个托管Agent,且不继承父上下文:和
twg-bench-lite-control。twg-bench-lite-test - 将完全相同的提示词作为完整用户消息发送给每个子Agent。请勿添加分支特定指令;分支隔离由托管Agent配置负责。测试Agent仅可将TWG路由计划作为工具选择策略接收,不可将其作为回答证据。
- 除非用户要求,否则请勿设置模型/推理覆盖参数;若设置,需对两个分支应用相同的覆盖。
- 等待两个子Agent完成,并记录它们的Agent ID或会话链接。
- 父会话无需再次解决任务,仅负责对比、评估和生成报告。
在父会话中打印生命周期进度:启动、提示词已捕获、Agent已启动、各分支完成、输出已读取、质量已评审、报告已编译、完成。
Quality
质量评估
Read both answers before interpreting token delta. Classify as ,
, , , or .
is a real outcome for missing, unusable, or non-comparable
answers, not a placeholder. Claim observed savings only for or
; for , report capability instead.
equivalenttwg-bettercontrol-bettercapability-gainnot-comparablenot-comparableequivalenttwg-bettercapability-gainEvaluate coverage, correctness, evidence quality, directness, and limitations.
Do not assign numeric quality scores.
在分析token增量前,先阅读两个回答。将结果分类为(等效)、(TWG更优)、(控制分支更优)、(能力提升)或(无法对比)。是针对缺失、不可用或无法对比的回答的真实结果,而非占位符。仅在结果为或时声明观测到的成本节省;若为,则报告能力提升情况。
equivalenttwg-bettercontrol-bettercapability-gainnot-comparablenot-comparableequivalenttwg-bettercapability-gain评估维度包括覆盖范围、正确性、证据质量、直接性和局限性。请勿分配数值质量分数。
Lite Report
精简报告
Return a compact report in the parent answer with:
- prompt text,
- control/test status and final outputs,
- observed tool families and TWG command evidence when available,
- token/tool/duration metrics when the agent runtime exposes them,
- quality classification and rationale,
- integrity notes: identical prompt, expected roles, TWG absent from control, TWG used in test, and any measurement gaps.
If token/session metrics are unavailable, say "not measured"; do not estimate.
Describe the result as one observed prompt comparison, not a universal
efficiency rate.
在父会话的回答中返回一份紧凑报告,包含:
- 提示词文本,
- 控制/测试分支的状态和最终输出,
- 观测到的工具家族及可用时的TWG命令证据,
- Agent运行时暴露的token/工具/时长指标,
- 质量分类及理由,
- 完整性说明:提示词完全相同、角色符合预期、控制分支未使用TWG、测试分支使用了TWG,以及任何测量缺口。
如果token/会话指标不可用,请注明“未测量”;请勿估算。将结果描述为单次观测到的提示词对比,而非通用效率比率。