How real projects brief Claude Code: the commands, conventions and warnings they put in CLAUDE.md.
K-Dense-AICLAUDE.mdK-Dense-AI/claude-scientific-writer/CLAUDE.md
verifiable citations. **Default Format:** LaTeX with BibTeX citations unless otherwise requested. **Quality Assurance:** Every PDF is automatically reviewed for formatting issues and iteratively improved until visually clean and professional. **CRITICAL
jhlee0409CLAUDE.mdjhlee0409/claude-code-history-viewer/CLAUDE.md
search - ✅ Web search results (`type: "web_search_tool_result"`) - ✅ Document content (`type: "document"`) - PDF and plain text - ✅ Search results (`type: "search_result"`) - ✅ MCP tool use (`type: "mcp_tool_use"`) - Model
AgriciDanielCLAUDE.mdAgriciDaniel/claude-blog/CLAUDE.md
category quality scoring (stdlib) scripts/blog_preflight.py # 5-gate delivery contract runner (v1.9.0) scripts/blog_render.py # md -> html -> pdf renderer; XSS-safe JSON-LD (v1.9.0) scripts/blog_hygiene.py # Optional deterministic hygiene: lazy-load imgs + auto
deusyuCLAUDE.mddeusyu/translate-book/CLAUDE.md
test_*.py' -v python3 -m compileall scripts tests ``` Then test with a small PDF to verify the full pipeline: ```bash python3 scripts/convert.py /path/to/small.pdf --olang zh # then run translation
doobidooCLAUDE.mddoobidoo/mcp-memory-service/CLAUDE.md
Chunker defaults: 1000 chars with 200 overlap. `semtools_loader.py` is optional and adds LlamaParse for PDF, DOCX and PPTX. ## Test Architecture About 2,400 tests across roughly 216 files. `tests/` mirrors
vstorm-coCLAUDE.mdvstorm-co/full-stack-ai-agent-template/CLAUDE.md
upload - Sync sources: configurable connectors (Google Drive, S3) with scheduled sync - 3 PDF parsers: PyMuPDF, LiteParse, LlamaParse (runtime selection via env var) - Image description via LLM vision API (optional
zzetCLAUDE.mdzzet/gortex/CLAUDE.md
drive the review itself. - **Multimodal + broad ingest** — image files (`KindImage` assets with format/dimensions/sha256) and PDF documents (per-page searchable `KindDoc` nodes) are graph nodes; new first-class extractors cover Terraform/HCL
KlotzketteCLAUDE.mdKlotzkette/claude-fuer-deutsches-recht/CLAUDE.md
jede Testakte gilt der unveränderte zweisprachige Hinweis aus [`AGENTS.md`](./AGENTS.md) und [`testakten/QUALITAETSSTANDARD.md`](./testakten/QUALITAETSSTANDARD.md). Gesamt-PDF und Einzel-PDFs enthalten weder eine Hinweisseite noch diesen Warntext. Stattdessen steht der vollständige Hinweis
Open-Source-LegalCLAUDE.mdOpen-Source-Legal/OpenContracts/CLAUDE.md
reactive vars, never touch URLs directly - Deep linking and canonical redirects handled automatically 3. **PDF Annotation System** (see `.cursor/rules/pdf-viewer-and-annotator-architecture.mdc`): - **Virtualized rendering**: Only visible pages (+overscan) rendered for performance - Binary search
fcakyonCLAUDE.mdfcakyon/claude-codex-settings/.claude/CLAUDE.md
haven't verified.** 1. **Author Names**: Verify exact author names from the actual paper PDF or official publication page. Do not guess or hallucinate author names based on similar-sounding
ClawBioCLAUDE.mdClawBio/ClawBio/CLAUDE.md
image segmentation, cell counting, segmentation mask | `skills/cell-detection/` | Run `cell_detection.py` | | WES clinical report English, exome PDF report, whole exome sequencing report, clinical exome PDF | `skills/wes-clinical-report-en/` | Run `wes_clinical_report_en.py` | | WES clinical report Spanish
917DhjCLAUDE.md917Dhj/DeepPaperNote/CLAUDE.md
examples: - `给这篇论文生成深度笔记` - `写一篇高质量论文精读笔记` - `把这篇文章整理成 obsidian 笔记` - `/deeppapernote <paper title, DOI, arXiv ID, or local PDF path
zeenie-aiCLAUDE.mdzeenie-ai/OpenCompany/CLAUDE.md
# OpenCompany - Claude Documentation ## Project Overview This is a React Flow-based workflow automation platform
borgheiCLAUDE.mdborghei/Claude-Skills/CLAUDE.md
calendar prep, investor update, pitch deck, weekly review | | **Documents** | [documents/CLAUDE.md](documents/CLAUDE.md) | 4 skills: docx, pdf, pptx, xlsx audit (stdlib-only OOXML parsing) | | **Vertical Advisors** | [vertical-advisors/CLAUDE.md](vertical-advisors/CLAUDE.md
borgheiCLAUDE.mdborghei/Claude-Skills/documents/CLAUDE.md
four Office formats Anthropic ships official skills for. Each toolkit reads OOXML or PDF directly using the Python standard library only. ## Skills in This Domain | Skill | Folder | Tool
borgheiCLAUDE.mdborghei/Claude-Skills/markdown-html/CLAUDE.md
front of it and a shared design-token layer underneath. Distinct from `documents/` (docx, pdf, pptx, xlsx), which *parses* existing OOXML and PDF binaries. This domain *produces* HTML from markdown
borgheiCLAUDE.mdborghei/Claude-Skills/vertical-advisors/CLAUDE.md
Investor-facing materials | `personal-productivity/investor-update-generator`, `pitch-deck-reviewer` | | Document handoff hygiene | `documents/` (docx, pdf, pptx, xlsx auditors) | | Board / executive coordination | `c-level-advisor/` skills + `agents/c-level/` agents | ## Adding
magicyuan876CLAUDE.mdmagicyuan876/mineru-tianshu/CLAUDE.md
Engine routing (`litserve_worker.py::_process_task`) Per-task pipeline: vLLM container switch → legacy Office conversion → PDF split → watermark removal → engine dispatch → normalize → persist. Dispatch is keyed on the task's `backend
tfriedelCLAUDE.mdtfriedel/claude-office-skills/CLAUDE.md
Repository Overview This is a skills repository for Office document manipulation (PPTX, DOCX, XLSX, PDF). Each skill provides workflows, scripts, and documentation for working with specific file formats. The repository
henrydaumCLAUDE.mdhenrydaum/second-brain/CLAUDE.md
extension list, which would drift from what is installed. That is right for a PDF and inverted for a pointer: `parse_gdoc` opens with `json.loads(sdk.fs.read(path))`, so being textual
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.
