paper-reach

分类: 调研与分析 | 上传者: Dai0-2 | 下载: 0 | 版本: v1.0（最新）

使用 Paper-Reach 进行证据意识文献筛选。搜索候选论文、保守地筛选摘要、查看可用的全文，并生成结构化 JSON。

更新日志: Source: GitHub https://github.com/Dai0-2/Paper_Reach

目录结构

当前层级: tree/main/

📁 .claude-plugin/
- 📄 marketplace.json 442 B
- 📄 plugin.json 662 B
📁 .github/
- 📁 ISSUE_TEMPLATE/
  - 📄 bug_report.md 482 B
  - 📄 feature_request.md 522 B
- 📁 workflows/
  - 📄 ci.yml 576 B
- 📄 pull_request_template.md 358 B
📁 agents/
- 📄 openai.yaml 440 B
📁 docs/
- 📄 agent-integration.md 5.4 KB
- 📄 architecture.md 2.4 KB
- 📄 browser-cookies.md 1.9 KB
- 📄 github-release-v0.1.0.md 1.5 KB
- 📄 install.md 3.5 KB
- 📄 publishing.md 2.8 KB
- 📄 release-notes-v0.1.0.md 1.7 KB
- 📄 roadmap.md 976 B
- 📄 usage.md 4.5 KB
📁 examples/
- 📁 agent-recipes/
  - 📁 codex/
    
    📄 query.json 688 B
    
    📄 README.md 1.4 KB
  - 📁 openclaw/
    
    📄 query.json 723 B
    
    📄 README.md 1.3 KB
  - 📄 README.md 674 B
- 📁 auth/
  - 📄 cookies.example.json 116 B
  - 📄 headers.example.json 81 B
- 📁 china_disaster_exposure/
  - 📄 query.json 724 B
  - 📄 README.md 1.1 KB
- 📄 README.md 316 B
📁 paper_reach/
- 📁 channels/
  - 📄 __init__.py 304 B
  - 📄 arxiv.py 3.4 KB
  - 📄 base.py 594 B
  - 📄 local_files.py 2.6 KB
  - 📄 mock_source.py 1.9 KB
  - 📄 openalex.py 3.5 KB
- 📁 fetchers/
  - 📄 __init__.py 165 B
  - 📄 base.py 567 B
  - 📄 open_access.py 5.1 KB
  - 📄 utils.py 4.3 KB
- 📁 parsers/
  - 📄 __init__.py 181 B
  - 📄 base.py 476 B
  - 📄 pymupdf_parser.py 742 B
  - 📄 simple_text_parser.py 502 B
- 📁 ranking/
  - 📄 __init__.py 86 B
  - 📄 profiles.py 981 B
  - 📄 rubric.py 17.9 KB
- 📁 schemas/
  - 📄 paper_result.schema.json 4.2 KB
  - 📄 query_input.schema.json 1.4 KB
- 📄 __init__.py 78 B
- 📄 cli.py 13.1 KB
- 📄 config.py 1008 B
- 📄 io_utils.py 1.0 KB
- 📄 models.py 6.3 KB
- 📄 retrieval.py 9.6 KB
- 📄 summarize.py 2.2 KB
- 📄 workflow.py 23.7 KB
📁 scripts/
- 📄 check-install.sh 1.0 KB
- 📄 sync.sh 1.2 KB
📁 skills/
- 📁 paper-ranker/
  - 📁 examples/
    
    📄 score_example.json 425 B
  - 📄 scoring.md 315 B
  - 📄 SKILL.md 2.1 KB
- 📁 paper-reader/
  - 📄 schema.json 399 B
  - 📄 SKILL.md 2.1 KB
- 📁 paper-search/
  - 📁 examples/
    
    📄 sample_input.json 612 B
    
    📄 sample_output.json 6.0 KB
  - 📄 rubric.md 312 B
  - 📄 SKILL.md 2.7 KB
📁 tests/
- 📁 fixtures/
  - 📄 sample_metadata.json 308 B
  - 📄 sample_paper.txt 205 B
- 📄 conftest.py 182 B
- 📄 test_fetchers.py 2.0 KB
- 📄 test_models.py 1.7 KB
- 📄 test_retrieval.py 3.4 KB
- 📄 test_rubric.py 5.7 KB
- 📄 test_summarize.py 5.3 KB
- 📄 test_workflow.py 5.9 KB
📄 .gitignore 164 B
📄 AGENTS.md 2.2 KB
📄 CHANGELOG.md 1.0 KB
📄 CONTRIBUTING.md 1.6 KB
📄 gemini-extension.json 686 B
📄 LICENSE 1.1 KB
📄 pyproject.toml 688 B
📄 README.md 8.1 KB
📄 README_EN.md 9.1 KB
📄 SKILL.md 3.0 KB

SKILL.md

---
name: paper-reach
version: "0.1.0"
description: "Run evidence-aware literature screening with Paper-Reach. Search candidate papers, screen abstracts conservatively, review full text when available, and emit structured JSON."
argument-hint: "screen topic query.json, review local PDFs, run literature screening for a research question"
allowed-tools: Bash, Read, Write
homepage: https://github.com/Dai0-2/paper_reach
repository: https://github.com/Dai0-2/paper_reach
author: Paper-Reach contributors
license: MIT
user-invocable: true
---

# Paper-Reach Skill

Use this skill when the user wants a rigorous literature-review workflow rather than ad hoc web search.

## Purpose

Paper-Reach is a reusable screening workflow for:

- candidate paper retrieval
- conservative abstract screening
- optional full-text fetching and review
- evidence extraction
- rubric-based ranking
- JSON-first outputs for literature review and gap analysis

## Host-Agnostic Workflow

1. Find the Paper-Reach installation root if you need repository files.
2. Read `AGENTS.md` for repository-wide expectations.
3. Use the relevant sub-skill under `skills/`:
   - `skills/paper-search/SKILL.md`
   - `skills/paper-reader/SKILL.md`
   - `skills/paper-ranker/SKILL.md`
4. Prefer the installed `paper-reach` CLI over reimplementing screening logic in prompts.
5. Consume the generated JSON and reason from explicit evidence and conservative decisions.

## Finding The Repository Root

If the current working directory is not obviously the repo root, search these common locations and use the first directory that contains both `AGENTS.md` and `paper_reach/cli.py`:

```bash
for dir in \
  "." \
  "${CLAUDE_PLUGIN_ROOT:-}" \
  "${GEMINI_EXTENSION_DIR:-}" \
  "$HOME/.claude/skills/paper-reach" \
  "$HOME/.agents/skills/paper-reach" \
  "$HOME/.codex/skills/paper-reach"; do
  [ -n "$dir" ] && [ -f "$dir/AGENTS.md" ] && [ -f "$dir/paper_reach/cli.py" ] && PAPER_REACH_ROOT="$dir" && break
done
```

If you only need execution and the package is already installed, you can skip repository path discovery and call the CLI directly.

## Execution Preference

Prefer these commands:

```bash
paper-reach doctor
paper-reach example-query
paper-reach screen --input query.json --output screen.json --high-recall --retrieval-limit 120
paper-reach review --input query.json --output review.json --local-path ./papers
paper-reach fetch-fulltext --input query.json --output review.json --download-dir ./downloads
paper-reach run --input query.json --output result.json
```

Fallback if the console script is unavailable:

```bash
python -m paper_reach.cli doctor
python -m paper_reach.cli screen --input query.json --output screen.json
```

## Guardrails

- Never treat title-only relevance as enough for final selection.
- Treat abstract-only support as coarse evidence.
- Keep unsupported papers as `ambiguous` or `need_fulltext`.
- Prefer JSON outputs over free-form prose when handing results back to the user or another agent step.
- If remote retrieval fails, fall back to offline mode or local files instead of inventing evidence.

登录后下载/点赞/收藏 ❤ 5 | ★ 0

paper-reach

目录结构

SKILL.md

举报内容

提示