agentleFS
XiaYiHannXiaYiHann/scholar-3/skills/research-proposal/SKILL.md

research-proposal

Skill0 starsChanged 9 months agoMITView on GitHub

进入 Proposal 模式(命令入口 `/proposal`)- 在单一冻结 Idea 下创建分阶段 Proposal(plan + tasks),并准备执行。

8 other projects use a version of this.

Adopted by

Most recently taken up by

The file

Files
SKILL.md

进入 Proposal 模式(本阶段首先是:创建/维护分阶段 proposal 工件)。

硬前置(默认严格)

  • 必须存在:
    • research/idea.md
    • research/idea.meta.yaml
    • research/evidence-ledger.md
  • 必须先读取 research/idea.md 中的 Proposal Handoff Capsule(若缺失,先回 /discover 补齐)
  • 若 research/idea.meta.yaml 的 state != frozen:
    • 默认:拒绝进入正式实验执行
    • 只能给两个选项:
      1. 回 /discover 补齐并 freeze
      2. 用户明确选择 “继续在 draft 下做探索性实验”(必须在 report 里标注为 exploratory)

目标(本阶段你必须输出的工件)

对一个 proposal 目录(research/proposals/Pxx-<slug>/)生成/更新:

  • proposal.md(意图、范围、目标 IDs、成功/停止条件)
  • experiment-plan.md(预注册 + E* 映射,禁止写结果)
  • tasks.md(checkbox 严格格式,apply 追踪)

模板来源:

  • scholar/schemas/scholar-research/templates/proposal.md
  • scholar/schemas/scholar-research/templates/experiment-plan.md
  • scholar/schemas/scholar-research/templates/tasks.md

入口问诊(AskUserQuestion-first)

建议问:

  1. 选择 proposal:新建 / 继续 Pxx / 查看已有列表
  2. 本 proposal 的目标:验证哪个 H/F,支持哪个 C
  3. 本次只做 planning 还是要进入执行(执行建议用 /apply)

Proposal 编号规则(强制)

  • 从 research/idea.meta.yaml 读取 proposal_counter
  • 新建 proposal 时:
    • proposal_counter += 1
    • 生成 P01/P02...
    • proposal slug 用 kebab-case(简短)

对照规则(不可违反)

  • proposal 的 Target IDs 必须来自 research/idea.md 的 H/F/C。
  • 禁止在 experiment-plan 里改 primary metric / tuning protocol(除非走 amendment)。
  • proposal.md 必须显式引用 handoff capsule 的:
    • locked metric/tuning
    • baseline set
    • top risks
    • open blockers

输出策略(默认)

  • 生成/更新 proposal 三件套后,建议用户运行:
    • /apply 执行 tasks
    • /verify 做审计与下一步决策

Download now (ZIP)

Does it work?

Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.

No reviews yet

People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.

  • It worked
  • Partly
  • It didn't
What did you use it for? e.g. Reviewing pull requests in a Rails app

Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.

Your trail

Files you open will collect here, so you can find them again.

Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.

Now do this for your own.

Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.