Sync ParaSkill rewrite-abstract-with-commitments

LICENSE

@@ -0,0 +1,21 @@
+MIT License
+
+Copyright (c) 2026 bhxdianzhang
+
+Permission is hereby granted, free of charge, to any person obtaining a copy
+of this software and associated documentation files (the "Software"), to deal
+in the Software without restriction, including without limitation the rights
+to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
+copies of the Software, and to permit persons to whom the Software is
+furnished to do so, subject to the following conditions:
+
+The above copyright notice and this permission notice shall be included in all
+copies or substantial portions of the Software.
+
+THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
+AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
+OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+SOFTWARE.

README.md

@@ -0,0 +1,56 @@
+---
+license: mit
+language:
+- zh
+- en
+tags:
+- codex-skill
+- ai-agent
+- scientific-writing
+- research-workflow
+- paraskill
+- paradoxgpt
+- writer
+pretty_name: RewriteAbstractWithCommitments
+---
+
+# RewriteAbstractWithCommitments
+
+`rewrite-abstract-with-commitments` is a ParaSkill research workflow skill for the `writer` family.
+
+## Purpose
+
+Rewrite a paper abstract so it makes clear, defensible commitments rather than vague promises or overclaims. Use when an abstract diagnosis already identifies weak realization, unclear contribution, unsupported claims, or poor claim calibration, and the user needs a revised abstract with rationale. Output is a revision plus sentence-level reasoning; it does not redesign the method or invent missing experiments.
+
+## Metadata
+
+- Family: `writer`
+- Owner model: `ParadoxGPT-Writer-4B`
+- Output type: `revision`
+- Version: `0.1`
+- Default language: `zh`
+
+## Included Files
+
+```text
+RewriteAbstractWithCommitments/
+  SKILL.md
+  agents/openai.yaml
+  README.md
+  LICENSE
+```
+
+`SKILL.md` is the executable skill instruction. `README.md` is the Hugging Face card.
+
+## Installation
+
+Copy this folder into a Codex-discoverable skills directory, or reference `SKILL.md` directly from an agent workflow.
+
+```bash
+mkdir -p ~/.codex/skills
+cp -r RewriteAbstractWithCommitments ~/.codex/skills/rewrite-abstract-with-commitments
+```
+
+## Source
+
+This skill is part of the ParaSkill library for ParadoxGPT. The full routing table is maintained in `SKILL_INDEX.json` in the source bundle.

SKILL.md

@@ -0,0 +1,125 @@
+---
+name: rewrite-abstract-with-commitments
+description: >-
+  Rewrite a paper abstract so it makes clear, defensible commitments rather
+  than vague promises or overclaims. Use when an abstract diagnosis already
+  identifies weak realization, unclear contribution, unsupported claims, or
+  poor claim calibration, and the user needs a revised abstract with rationale.
+  Output is a revision plus sentence-level reasoning; it does not redesign the
+  method or invent missing experiments.
+metadata:
+  owner_model: ParadoxGPT-Writer-4B
+  skill_family: writer
+  version: "0.1"
+  output_type: revision
+  default_language: zh
+  trigger_keywords: [abstract, rewrite, commitment, overclaim, claim calibration, 摘要, 改写, 承诺]
+  required_inputs: [old_abstract, paper_context]
+  optional_inputs: [diagnosis, target_realization, claim_constraints, target_venue]
+  handoff_to: [diagnose-aha-moment, detect-overclaim-missing-closure, build-intro-argument-chain]
+---
+
+# Rewrite Abstract With Commitments
+
+## Purpose
+
+把 abstract 改成一个可审稿的承诺集合：清楚说明问题、关键 gap、核心 insight、方法 move、证据范围和贡献边界。这个 skill 不是润色器，而是帮 Writer 把 abstract 中的每一句都变成有后文闭合责任的 claim。
+
+## When to Use
+
+- 已有 abstract，但读起来像背景罗列、贡献不清或 insight 被埋住。
+- `diagnose-aha-moment` 已指出 central realization、commitment 或 closure 问题。
+- 需要把过大的 claim 降到实验能支撑的范围内。
+
+## Do Not Use
+
+- 还不知道论文的核心 realization → 先用 `diagnose-aha-moment`。
+- 主要问题是 claim 没实验支撑 → 先用 `detect-overclaim-missing-closure` 或 `map-claims-to-experiments`。
+- 要规划 introduction 段落链 → 用 `build-intro-argument-chain`。
+
+## Required Inputs
+
+- `old_abstract`: 原 abstract。
+- `paper_context`: 至少包含 title、problem、method_summary、experiment_summary。
+- `diagnosis`: optional，来自 `diagnose-aha-moment` 的诊断。
+- `target_realization`: optional，希望审稿人获得的核心认知更新。
+- `claim_constraints`: optional，哪些 claim 必须保守、哪些不能写。
+
+## Output Contract
+
+输出必须至少包含：
+
+1. **revised_abstract** - 一版完整 abstract。
+2. **sentence_by_sentence_rationale** - 每句话承担什么承诺，为什么这样写。
+3. **claim_calibration_notes** - 哪些 claim 被加强、弱化或限定范围。
+4. **remaining_risks** - 仍可能被 reviewer 批的点。
+5. **handoff_suggestion** - 如果还有实验/逻辑缺口，应转给哪个 skill。
+
+## Procedure
+
+1. 从 paper_context 中抽取 problem、gap、core insight、method move、evidence scope。
+2. 对 old_abstract 标注每一句的功能：背景、gap、insight、method、evidence、claim。
+3. 删除或压缩不能服务核心 realization 的背景句。
+4. 重写 abstract，使贡献边界和实验支撑范围一致。
+5. 逐句说明承诺，指出哪些地方需要后文闭合。
+
+## Quality Bar
+
+一个好的输出必须：
+
+- revised_abstract 能让人读出一个清楚的 reviewer realization。
+- 每个强 claim 都能在 method 或 experiment 中找到闭合路径。
+- 不为了显得强而写成普遍结论；范围限定具体。
+- rationale 能解释为什么这么改，而不是只说“更清晰”。
+
+一个差的输出：
+
+- 只换同义词，没有改变承诺结构。
+- 添加 paper_context 没有提供的新结果。
+- 把 limitation 藏起来，导致后续 reviewer 更容易抓 overclaim。
+
+## Failure Modes
+
+- **润色替代重构**：只改句子流畅度，不修 claim。
+- **新发明贡献**：为了好看添加论文没有做的东西。
+- **过度保守**：把真正有价值的 insight 弱化成普通工程改进。
+- **忽略证据范围**：实验只在小范围成立，abstract 写成通用规律。
+
+## Handoff
+
+- 如果 revised_abstract 仍缺核心 realization → `diagnose-aha-moment`
+- 如果发现关键 claim 没有实验闭合 → `detect-overclaim-missing-closure`
+- 如果 abstract 改完后需要规划 introduction 链条 → `build-intro-argument-chain`
+
+## Example
+
+Input:
+
+```
+old_abstract: <背景很长，最后才说 interaction scaling>
+paper_context: <long-horizon agents, method, experiments>
+diagnosis: insight 被埋住，C2 机制简单性证据不足
+```
+
+Output:
+
+```
+revised_abstract:
+  Long-horizon agents are often improved by stronger single-step policies, but
+  this leaves interaction itself under-modeled as a scaling resource. ...
+
+sentence_by_sentence_rationale:
+  S1: 建立 gap - 现有工作偏单步能力。
+  S2: 给出 core insight - interaction 是可 scaling 的资源。
+  S3: 方法 move - 提出简单交互扩展机制。
+  S4: 证据范围 - 在长程任务族上观察持续提升。
+
+claim_calibration_notes:
+  - 把“universally improves”降为“on long-horizon task families studied here”。
+  - 不再声称机制完全解释增益���只写“suggests”。
+
+remaining_risks:
+  - 机制证据仍偏弱，应补 ablation 或在正文降 claim。
+
+handoff_suggestion: detect-overclaim-missing-closure
+```

agents/openai.yaml

@@ -0,0 +1,4 @@
+interface:
+  display_name: "Rewrite Abstract With Commitments"
+  short_description: "Rewrite a paper abstract so it makes clear, defensible commitments rather than vague promi"
+  default_prompt: "Use $rewrite-abstract-with-commitments for this research-writing task."