Sluice
Franlinozz/Sluice/CLAUDE.md
dark-bg vs light-bg), then rename/organize. 19. WHITEPAPER VERIFICATION LOOP: after generating the PDF, rasterize EVERY page (pdftoppm -png) and visually inspect each image. Blank/broken pages = regenerate
How real projects brief Claude Code: the commands, conventions and warnings they put in CLAUDE.md.
Franlinozz/Sluice/CLAUDE.md
dark-bg vs light-bg), then rename/organize. 19. WHITEPAPER VERIFICATION LOOP: after generating the PDF, rasterize EVERY page (pdftoppm -png) and visually inspect each image. Blank/broken pages = regenerate
FutureRootsDE/legal-audit-de/CLAUDE.md
Codebase → ` /docs/legal-audit/` | | `/legal-audit --compare` | Audit + Diff-Report gegen letztes Audit aus `audits/`-Archiv | | `/legal-audit --pdf` | Audit + generiert `anwalts-briefing.html` (+ PDF wenn pandoc vorhanden) | | `/legal-audit-live ` | Live-Browser-Check via chrome-devtools
maweiruc/proofread-stat-paper/CLAUDE.md
intentionally planted errors) 2. Verify `examples/output/test_paper_annotated.tex` compiles: `pdflatex -interaction=nonstopmode test_paper_annotated.tex` 3. Check the PDF for red (grammar) and blue (technical) annotations 4. Confirm `proofread_report.md` captures findings with correct categories/severity
myths-labs/muse/CLAUDE.md
before-completion`** | | **GEO/SEO optimization** | **`geo-seo`** → `geo-audit` / `geo-citability` / `geo-schema` / `geo-report-pdf` | | **Git 安全守卫** | **`git-security-guard`**(commit/push 前检查泄露) | | **去 AI 味** | **`deai-humanizer`**(AI 文字 → 自然表达
myths-labs/muse/templates/CLAUDE.md
before-completion`** | | **GEO/SEO optimization** | **`geo-seo`** → `geo-audit` / `geo-citability` / `geo-schema` / `geo-report-pdf` | | **Web Deck/PPT** | **`frontend-slides`** (zero-dep HTML slides + PPT conversion + 12 presets + Vercel deploy + PDF
zoharbabin/google-researcher-mcp/CLAUDE.md
development) ## Architecture ``` src/ ├── server.ts # Main server entry point ├── cache/ # Caching layer with persistence ├── documents/ # PDF, DOCX, PPTX parsing ├── prompts/ # MCP Prompts primitive ├── resources/ # MCP Resources primitive ├── schemas/ # Output schemas ├── shared
Johnson-Jia/video-clipforge/CLAUDE.md
编排和模式选择的权威工作流定义。 ## 什么是 ClipForge ClipForge 是 AI 驱动的通用短视频制作系统。通过 DAG 编排管线,将任意内容(文本、URL、PDF、GitHub 数据等)转化为抖音就绪的竖屏视频,含旁白、BGM 和封面。内容分类(GitHub、漫画、小说等)通过 `clipforge/categories/` 中的分类配置定义。 **核心管线**:`内容 → 设计
xu-xiang/awesome-top-skills/tags/claude.md
# Tag: claude - Back: [Tags](README.md) - Tagged skills: 219 | Skill | Repo | Summary | ⭐ | ⬇️ | Updated | Tags | | - | - | - | -: | -: | - | - | | [product
gokeshenzhen/awesome-formal-verification-skill/CLAUDE.md
split a module that has genuinely outgrown a flat file. ### Source ingestion — figures in PDF and Office docs (build-time) Source documents carry their highest-value content (SVA/RTL/Tcl snippets, waveforms
Joanium/Joanium-Classic/CLAUDE.md
with `electron-updater` for auto-updates - **Notable deps:** xterm.js (terminal), exceljs, jszip, mammoth, pdf-parse --- ## Workspace Structure ``` Core/Electron/ Main-process entry + window management Packages/Capabilities/ Integration modules (APIs, system tools) Packages/Features
marcosmarf27/pje-ia/CLAUDE.md
Anthropic nunca entra num request Gemini (e vice-versa; `montarBlocos`/`subirPecas` conferem). PDF Gemini: ≤ 50 MB/1000 págs., 258 tokens/pág. Upload é resumable + poll de `state:ACTIVE`. - **Sem citações
oomkapwn/enquire-mcp/CLAUDE.md
overclaim instance #13 (v3.9.0-rc.4): CLAUDE.md header line said "**Still deferred to v3.9.0+:** ... OCR'd PDF watcher embed-sync, HNSW in-memory live update, R-10 adaptive refill" — but the status
arcadeai-labs/agent-library/CLAUDE.md
multi-modal assets (text, code, images, PDFs) ### Current Status - **Multi-Modal Support Complete:** Code, PDF, and image parsing with asset type preservation - **Parser Registry:** Automatic parser selection based on file
BulloRosso/etienne/CLAUDE.md
Project-specific Claude Code settings ├── data/ │ └── permissions.json # Tool permissions └── out/ # Output files ``` ## Working with PDF and Office Format Documents The standard skill **office-and-pdf-documents** provides document parsing capabilities
joneqian/claude-skills-suite/CLAUDE.md
skills-suite/ ├── skill_seekers/ # Main Python package │ ├── cli/ # CLI tools (doc_scraper, github_scraper, pdf_scraper, etc.) │ └── mcp/ # MCP server implementation (FastMCP-based) ├── configs/ # Pre-built scraping configs for popular
reichenbach/iwork_mcp/CLAUDE.md
down first, then write new data at row 0. - **JXA export format strings**: Use `"PDF"` not `"Numbers PDF"`/`"Pages PDF"`/`"Keynote PDF"`. The app-prefixed strings cause
nicholasglazer/gnosis-mcp/CLAUDE.md
extras: `[postgres]` (asyncpg), `[embeddings]` (onnxruntime, tokenizers, numpy, sqlite-vec), `[web]` (httpx, trafilatura), `[rst]` (docutils), `[pdf]` (pypdf), `[formats]` (docutils + pypdf). Model download uses stdlib `urllib` (no `huggingface-hub` dependency). ## Tools ### Read
estevesjh/mcp-ads-arxiv/CLAUDE.md
reading full-text files before aligning with the user**. **IRON RULE: never read a PDF's raw bytes or text. Always serve the docling-generated markdown** produced by `get_paper
crazynomad/skills/CLAUDE.md
APIs; `doc-mindmap` is cross-platform) ## Available Skills | Plugin | Skill | Script | Dependencies | |--------|-------|--------|--------------| | media-skills | pdf-to-images | `scripts/pdf_to_images.py` | ImageMagick (`brew install imagemagick`) | | media-skills | podcast-downloader | `scripts/download_podcast.py` | `pip install requests
ractive/hyalo/.claude/CLAUDE.md
mailto:`, `file://`) | `attachment` (resolved to a non-`.md` vault file: an image, a PDF, an Obsidian `.base`). `external` and `attachment` links are never broken and are not graph edges
A file Claude Code reads at the start of every session. It holds the commands, conventions and warnings the agent needs for this project.
At the repository root. Claude Code also reads CLAUDE.md files in subdirectories when it works there.
Build and test commands, the project's layout, conventions that aren't obvious from the code, and mistakes to avoid. Short files tend to work better than long ones.
Claude Code reads CLAUDE.md; most other agents read AGENTS.md. Many projects keep one and point the other at it.