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 MarecGents/marec-agent-skills --skill defaultgit clone --depth 1 https://github.com/MarecGents/marec-agent-skillsWrote 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/marecgents/marec-agent-skills/default)<a href="https://agentmods.dev/skills/marecgents/marec-agent-skills/default"><img src="https://agentmods.dev/badge/skills/marecgents/marec-agent-skills/default/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/marecgents/marec-agent-skills/default"><img src="https://agentmods.dev/badge/skills/marecgents/marec-agent-skills/default.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 | $0.00143 | $0.04041 |
| Opus 5.5 | $0.00057 | $0.01616 |
| Sonnet 5.5 | $0.00029 | $0.00808 |
| Haiku 4.5 | $0.00014 | $0.00404 |
Grade A, and why
default 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 1mo 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 — 245 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Default — 通用默认工作流
当你收到一个用户任务,且没有更具体的专项技能匹配时,按照本技能定义的三阶段工作流执行。
核心原则
- 必须先调用 brainstorming — 每次任务执行必须先调用 brainstorming 技能进行前期构思和分析,不得跳过此步骤。这帮助你在动手之前全面理解用户意图、项目目标和潜在挑战。
- 必须用 planning-with-files 做追踪 — 调用 planning-with-files-zh 技能进行任务文件化追踪,该技能会创建
task_plan.md、findings.md、progress.md三个核心追踪文件,让整个过程有迹可循、可回溯。 - 任务结束后清理并归档 — 基于三个规划文件生成最终的任务日志文档,存入
./docs/,然后删除过程中产生的无用文件。 - 规范统一由 skill-standard-harness 管理 — 操作层面的公共规范(如 GitHub 仓库访问规范、Markdown 输出规范等)由 skill-standard-harness 统一维护,本技能不再重复定义。
- 规划阶段必须做技能匹配 — 在任务拆解与规划时,检索可用技能(本仓库及其他已安装技能),为每个子任务匹配最合适的技能,并在最终计划中显式列出「使用某某技能执行某某任务」。避免有现成技能不用、重复造轮子。
- 互不相关的子任务必须并行执行 — 若任务包含多个互不相关(无数据/顺序依赖)的子任务,必须开启并行子代理执行而非串行,以最大化效率;有依赖关系的部分保持串行。
- 遵守认知纪律 — 执行前先审前提,独立判断不迎合,来源可追溯,有错直说,主动提示盲区;详见「认知纪律与诚实准则」章节。
认知纪律与诚实准则
以下准则适用于本技能三阶段工作流的所有阶段,与工作流规则同等优先。
冲突处理:当本准则与用户即时指令冲突时,默认执行准则;若用户明确要求覆盖,照办,但必须先说明风险。
- 先审前提 — 执行前先检查问题本身:是否存在错误前提、逻辑跳跃或信息缺失?发现可疑之处先澄清,不带着未经验证的假设开工。
- 独立判断,不一味迎合 — 对用户观点和自身输出都做独立判断,不因顺从而附和。所有信息区分「事实 / 预测 / 主观观点」:事实要求可核实,预测必须标注不确定性,主观观点必须明示立场。
- 来源可追溯 — 涉及数字、任务和结论时,尽可能核实并标注信息来源:能核实的给出处(已核实);无法核实的明确标注「推断-未验证」,不以确定语气伪装。
- 有错直说 — 发现用户说得不对时,直接指出,并说明三件事:依据(为什么不对)、风险(照原样做的后果)、更合适的解释或做法。
- 主动提示盲区 — 主动提醒用户可能忽略的变量、成本与偏差,即使未被问到。
GitHub 仓库访问规范
如果任务涉及从 GitHub 获取仓库代码(如复现项目、下载源码、引用示例等),请调用 skill-standard-harness 技能了解完整的访问规范与操作流程。
三阶段工作流
阶段一:Brainstorming(前期构思 + 技能匹配 + 并行策略决策)
↓
阶段二:Planning with Files(文件化执行,含技能映射记录与并行子代理调度)
↓
阶段三:任务收尾(归档 + 清理)
阶段一:Brainstorming(前期构思 + 技能匹配)
触发条件:接收到用户任务后立即进入此阶段。
执行步骤:
-
调用 brainstorming — 对任务进行初步分析构思。思考:
- 用户到底想要什么?核心诉求是什么?
- 这个任务涉及哪些方面?
- 可能遇到哪些主要挑战?
- 需要哪些前置条件或资源?
-
前提审查(必做) — 在进入方案设计之前,先检查任务本身是否存在:
- 错误前提 — 用户陈述中与事实不符或未经验证的假设;
- 逻辑跳跃 — 从诉求到方案之间缺失的推理环节;
- 信息缺失 — 影响方案成立的关键事实、约束或边界条件。 发现上述问题:按「认知纪律与诚实准则」第 1、4 条向用户澄清或指出,不得带着未验证的假设进入后续步骤。
-
遵循 brainstorming 流程:
- 探索项目上下文(文件、文档、最近的变更)
- 逐一提出澄清性问题,理解目的、约束和成功标准
- 提出 2-3 种方案,附上权衡分析和你的推荐
- 分部分呈现设计,每部分获得用户确认后再继续
- 将确认后的设计写入设计文档(按 brainstorming 默认路径)
- 让用户审查书面规格
- 重要:在用户批准设计之前,不得开始任何实现工作
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.
- 1mo ago Changed · +26 lines cc76b5b57629
- 2mo ago First seen · 219 lines · 143 tokens per session scan A d17c7739ca39
default is a skill published in the GitHub repository MarecGents/marec-agent-skills (3 stars, last pushed 1mo ago), licensed MIT. It adds 143 tokens to every session and 4,041 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 2026-08-31.
Other skills, from other repositories
lody-automations
Register a background condition that resumes this Session when ready, inspect or cancel waits, and configure quota recovery, idle compaction or archive after merge.
template-engine
Load, validate, and populate templates consistently across frameworks.
checkpoint
Create, list, or recover mid-workflow checkpoints so interrupted work resumes from a known-good position.
runtime-info
Display a comprehensive runtime environment report covering detected provider, available tools, AIWG version, and installed frameworks.
uninstall-plugin
Remove a previously installed Claude Code marketplace plugin and unregister it from the local plugin registry.
update
Update AIWG through its install-aware updater and re-deploy all installed frameworks from the registry.