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 commands/alvis-haoh/gkd/workflowgit clone --depth 1 https://github.com/alvis-HaoH/gkdWrote 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/commands/alvis-haoh/gkd/workflow)<a href="https://agentmods.dev/commands/alvis-haoh/gkd/workflow"><img src="https://agentmods.dev/badge/commands/alvis-haoh/gkd/workflow.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 | $0.00050 | $0.01287 |
| Opus 5 | $0.00025 | $0.00643 |
| Sonnet 5 | $0.00010 | $0.00257 |
| Haiku 4.5 | $0.00005 | $0.00129 |
Grade A, and why
workflow scanned grade A with 0 findings against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured 3d ago.
A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.
Nothing flagged
None of the 26 patterns this scan looks for appear in this file: no shell pipes, no recursive deletes, no credential paths, no hidden text, no instruction-override or anti-refusal phrasing, no agent-config snooping. That is not a guarantee, it is the absence of the things that are checkable.
How it starts
The opening of the file, as written. The whole thing — 48 lines — stays where its author put it; the contents beside it link to each section on GitHub.
把用户的批量委派需求编排成 dynamic workflow,每个 item 一个换脑子进程并行处理。 批量委派的意义就是别让主上下文承担 N 个文件的开销:你拿到要处理的路径列表、编排好,读写实活交给子进程自己。
原始参数:$ARGUMENTS
怎么做
-
识别 items 来源 + 任务模板:从用户描述里找出 items(glob、目录、git diff 列表、明确列表)和"对每个 item 做什么"。描述含糊就用
AskUserQuestion问清。用 Glob/Bash(git:*) 拿路径列表(item 数 ≤ 200 通常 OK,过多问用户是否分批)。 -
选模型(默认从简):
- 用户显式传
--<modelKey>→ 全批锁定该模型。 - 同构批量(N 个文件相同改动)→ 默认统一一个便宜的能胜任模型最简单稳妥,不必逐 item 折腾。
- item 明显异构(部分含图片、部分高难、部分普通改写)才按 item 分模型,别在脚本里写死模型名。含图片的 item 只能派给支持视觉的模型:!
node "${CLAUDE_PLUGIN_ROOT}/scripts/gkd-runtime.mjs" --list-vision - worker/verifier 角色分工:若任务是"改 + 验"两步,让 worker(便宜模型干活)和 verifier(另一模型把关/换视角)走不同模型是有价值的。
- 用户显式传
-
写 workflow 脚本。注意 Workflow 脚本运行时没有
bash()hook,也不能直接跑 shell;agent()的 model 参数又只认 Claude 系标识符,换不了第三方模型脑。所以换脑只能这样链接:agent()起一个轻量启动器子代理(用便宜的haiku,活只是跑一条命令拿回 JSON),让它用 Bash 工具调gkd-runtime.mjs——真正换脑发生在 runtime spawn 的子进程那层(三方模型在那里干活、token 隔离)。核心 stage 形如(
pickModel按上面策略算 runtime 的--<modelKey>,不是写死;--write用户传了才加):const results = await pipeline( items, item => agent( `用 Bash 工具执行这条命令,把它的 stdout 原样作为你的最终输出返回,不要加任何解释:\n` + `node "${CLAUDE_PLUGIN_ROOT}/scripts/gkd-runtime.mjs" ${pickModel(item)} --json "<任务模板里的 {{item}} 替换成实际路径>"`, { model: 'haiku', label: `gkd:${item}` } ).then(out => JSON.parse(out)) )启动器用 haiku或sonnet 是因为它只是中转(spawn + 回传),不该烧贵模型;干实活的模型在 runtime 子进程里。并发度用 Workflow tool 默认,不用手设。
长任务超时:runtime 对每个委派子进程有默认 30 分钟超时兜底,到点杀掉归一成失败。批量里单个 item 预计会超 30 分钟(逐文件长生成、重活改动)时,把环境变量拼在命令最前面提高/关闭上限,否则该 item 会在 30 分钟被杀、记失败:
GKD_TIMEOUT_MS=3600000 node "${CLAUDE_PLUGIN_ROOT}/scripts/gkd-runtime.mjs" ...(1 小时),或GKD_TIMEOUT_MS=0(禁用)。注意这是单个子进程的时限,不是整个 workflow 的;常规同构小批量用默认即可。思考强度
--effort(none/low/medium/high/xhigh/max,claude/codex 通用):直接拼在 runtime 命令里即可,支持三种粒度——① 整批统一:每条命令都带同一个--effort xhigh;② 按 item:pickModel那样另写个pickEffort(item)按难度返回不同档(难 item 用 max、普通用默认不带);③ 按角色:worker stage 不带(省钱)、verifier stage 带--effort high(把关更严)。用户说"都仔细想""难的那批深想""验证环节用 high"之类就据此拼。 -
汇报:总数/成功/失败,失败 item 连错误原文列出。成功产出已落盘(--write)或在 result 里,不必全贴。
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 · 48 lines · 50 tokens per session scan A d131e013c889
workflow is a command published in the GitHub repository alvis-HaoH/gkd (10 stars, last pushed 1mo ago), licensed MIT. It adds 50 tokens to every session and 1,287 once invoked, about $0.0003 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 2026-08-31.
Other commands, from other repositories
OPSX: Fast Forward
Create a change and generate all artifacts needed for implementation in one go.
OPSX: Propose
Propose a new change - create it and generate all artifacts in one step.
OPSX: Onboard
Guided onboarding - walk through a complete OpenSpec workflow cycle with narration.
OPSX: Sync
Sync delta specs from a change to main specs.
OPSX: Apply
Implement tasks from an OpenSpec change (Experimental).
OPSX: Archive
Archive a completed change in the experimental workflow.