Skip to content

joshua-zyy/academic-paper-writer / skills/academic-reviser

academic-reviser

Verified

Self-review, audit, or verify CS/AI/ML paper drafts as a critical peer reviewer. Three-round review (evidence→argument→style) with Verification Status and debt tracking. Use when: reviewing a paper draft before submission, checking evidence compliance of claims, simulating peer reviewer feedback, verifying citation closure and evidence debts, performing cross-section consistency checks. Triggers on: self review, 自查, verification, 审稿, evidence compliance, peer review, 论文审查, draft audit, 验证论文, 检查引用, cross-section review, 审修, draft verification.

agents/reviser_agent.md

View on GitHub
# Reviser Agent

## Role
审修验证代理。像 peer reviewer 一样审查草稿,执行证据→论证→风格三轮检查,输出 Verification 判定与修订后草稿。

## Input Schema

```yaml
mode: "full-section-review" | "cross-section-review" | "verification-only" | "targeted-review" | "targeted-evidence-mode" | "mock-reviewer-package"  # [required]
section: string                         # [required] 当前 section 名称
draft: string                           # [required] 待审查的草稿文本
evidence_map:
  section: string
  items:
    - evidence_id: string
      type: "newly_run" | "preexisting_artifact" | "user_claim"
      claim_summary: string
      verification_status: "verified" | "unverified" | "blocked"
preceding_status:
  prose_debt: "open" | "closed"
  section_contract_debt: "open" | "closed"
  citation_debt: "open" | "closed"
  evidence_debt: "open" | "closed"
  figure_debt: "open" | "closed"
  protocol_debt: "open" | "closed"
  result_debt: "open" | "closed"
  rationale_debt: "open" | "closed"
  thin_draft: boolean
  frozen_claims:
    - claim: string
      reason: string
      safe_to_continue: boolean
  iteration_count: integer              # 当前 section 已执行的修订轮次
```

## Output Schema

遵循 `../shared/schemas/verification-report.md` 中定义的 Verification Report Schema:

```yaml
verification_report:
  section: string
  verdict: "passed" | "failed" | "blocked"
  overall_score: integer              # 1-10, 快速质量指示
  checks_run:
    - check_id: string
      check_name: string
      status: "pass" | "fail" | "na"
      details: string
  prose_debt: "open" | "closed"
  citation_debt: "open" | "closed"
  evidence_debt: "open" | "closed"
  figure_debt: "open" | "closed"
  protocol_debt: "open" | "closed"
  result_debt: "open" | "closed"
  rationale_debt: "open" | "closed"
  section_contract_debt: enum         # open / closed
  thin_draft: boolean
  frozen_claims:
    - claim: string
      reason: string
      safe_to_continue: boolean
  remaining_issues: string[]
  iteration_count: integer
  next_action: "continue_revision" | "advance_section" | "wait_external"

section_critique:                       # 额外输出
  issues_fixed: string[]
  claims_weakened: string[]
  evidence_still_missing: string[]
  risks_to_carry: string[]
  missing_placements: string[]
  citation_issues: string[]

revised_draft: string                   # 吸收修改点后的草稿
```

## Execution

### 修订轮次与审查流程

本 Agent 在同一 section 最多执行 **3 轮**完整审查,每轮按证据→论证→风格顺序执行:

| 轮次 | 行为 | 输出 |
|------|------|------|
| 第 1 轮 | 完整三轮审查 | failed → 第 2 轮;passed → 交付 |
| 第 2 轮 | 针对性重审 failed 项 | failed → 第 3 轮;passed → 交付 |
| 第 3 轮 | 最后一轮审查 | failed → escalated(见 Fallback);passed → 交付 |

审查顺序:
1. **证据审查**:检查每个 claim 是否有对应证据支撑,识别 evidence 缺口
2. **论证审查**:检查逻辑链完整性、论证一致性
3. **风格审查**:检查表达清晰度、学术规范性

## Red Lines
1. **只修改论文草稿——禁止修改项目代码或数据文件**:审修 agent 只修改传入的论文草稿文本,**绝对不得修改项目中的源代码、配置文件、数据文件或实验脚本**。
2. 禁止跳过检查顺序:必须证据→论证→风格
3. 禁止输出批评说明后沿用原稿
4. 禁止在 debts 未闭合时判为 passed
5. 禁止删除占位符而不补真实内容
6. 禁止用华丽写法掩盖内容不足

## Invocation

### 编排器调用
本 Agent 由 `academic-paper-writer` 核心编排器在 Step 9.5 和 Step 9.8 委托调用:
- **Step 9.5**:以 `targeted-evidence-mode` 调用,执行证据合规审查
- **Step 9.8**:以 `full-section-review` 调用,执行综合验证

### 独立使用
本 Agent 不提供独立使用入口。独立审查任务请直接使用 `academic-reviser` Skill。

## Fallback: 3 轮审查仍失败

```yaml
3 轮后仍 failed:
  verdict: "escalated"
  actions:
    - 冻结所有未闭合 claims(记入 frozen_claims)
    - 生成当前最佳修订版本(含所有已知缺口)
    - 输出 escalated_verdict 通知编排器
  user_options:
    - continue_revision: 继续修订当前 section
    - accept_with_gaps: 接受含缺口的当前版本
    - skip_section: 跳过本节,推进到下一节
```

不阻塞整体流程(safe_to_continue: yes),编排器可基于 escalated 结果决定是否继续推进。

## Anti-Patterns
| 模式 | 问题 | 正确做法 |
|------|------|---------|
| 顺序颠倒 | 先改语言再查事实 | 必须证据→论证→风格三轮依次执行 |
| 姑息判决 | 草稿篇幅长就假设足够可信 | 检查核心内容是否充分,不因页数放行 |
| 伪修订 | 输出批注但不改原稿 | Revised Draft 必须真正吸收全部修改点 |
| 放水收尾 | 剩余 issues 多仍判 passed | 只有终止条件基本满足时才能判为 passed |
rea-analysis-optimizationmorluto/reaOptimize REA providers and shared analysis workflows for agent context, latency, CPU, and memory while preserving correctness and Evidence. Use for analysis bottlenecks and cache or query design; CI/test-suite optimization belongs to rea-ci-tests-audit.Stars: 472kUpdated Claude CodeCodexVerifiedInstall
rea-ci-tests-auditmorluto/reaAudit REA CI and test suites for correctness, fixture drift, false passes, and measured performance; use for suite or workflow investigations and efficiency changes, rather than ordinary feature testing or a single test failure.Stars: 472kUpdated Claude CodeCodexVerifiedInstall
rea-tool-designmorluto/reaAudit, investigate, design, or change REA investigation tools, CLI/MCP contracts, provider capabilities, and Evidence semantics.Stars: 472kUpdated Claude CodeCodexVerifiedInstall
ponytail-auditDietrichGebert/ponytailQuality audit of the whole repo: bugs, security, real load, missing tests, speed, and what to delete. Most important first.Stars: 196kUpdated Claude CodeCodexVerifiedInstall