How real projects brief Codex, Cursor and every other agent that reads AGENTS.md.
JasonColapietroAGENTS.mdJasonColapietro/suede-creator-skills/AGENTS.md
every stale count from the live measurement npm run build:book # regenerates BOOK.md, the PDF and docs/book from those sources npm run build:cards # regenerates the per-skill CARD.md release
NeoTheCaptAGENTS.mdNeoTheCapt/RedteamAgent/agent/AGENTS.md
Board visit and fresh `/api/Challenges` fetch. - `Upload Type`: after `/file-upload` acceptance, submit a non-PDF/non-ZIP payload and check the consumer/scoreboard result instead of stopping at ZIP acceptance. - `User Credentials
Prism-ShadowAGENTS.mdPrism-Shadow/deepseek-harness-book/AGENTS.md
裁剪不得劈开文字或界面容器，也不得把过细的局部异常放大。 - 截图和示例中不得出现 API Key、访问令牌、个人路径、内部地址或其他敏感信息。 - dsh 的功能、配置和界面行为以当前源码或实际运行结果为准。 - 修改目录、标题、图片或 PDF 主题后，运行 `./scripts/build_typst_pdf.sh` 并检查生成的 PDF。 ## 仓库写作 Skills - 新写或大幅改写中文书稿时，优先使用 `.agents/skills/human-writing/`。 - 已有稿件需要降低 AI 写作痕迹时，使用 `.agents/skills/humanizer-zh
songzhuozhuAGENTS.mdsongzhuozhu/obsidian-llm-wiki/inspool-wiki-en/AGENTS.md
Stores raw source material such as web clips, articles, book excerpts, meeting notes, PDF transcripts, screenshots, and similar inputs. - This is the factual source layer. - The body is read-only
songzhuozhuAGENTS.mdsongzhuozhu/obsidian-llm-wiki/inspool-wiki-zh/AGENTS.md
原始资料可以保留原语言，不需要强制翻译；但 wiki 沉淀层的说明和结构应默认使用中文。 ## 三层结构 ### 1. `raw/` - 存放原始资料，例如网页剪藏、文章、书摘、会议纪要、PDF 转写稿、截图说明等。 - 这是事实来源层。 - 正文默认只读。 - 默认允许两类流程性变更：移动文件路径，以及更新少量 frontmatter 处理元数据。 - 除非用户明确要求，否则不得改写、删减或重写 `raw/` 内正文内容。 ### 2. `wiki
SpillwaveSolutionsAGENTS.mdSpillwaveSolutions/agent-brain/agent-brain-plugin/AGENTS.md
local, project-scoped memory of the docs and code it works on: - Index Markdown, PDF, and source code into one corpus per project (`.agent-brain/`). - Search with BM25, vector, hybrid
AyuilosAGENTS.mdAyuilos/Miffan/AGENTS.md
OpenAI, Google, Anthropic) - **common**: Common utilities and extensions - **document**: Document parsing module for handling PDF, DOCX, PPTX, and EPUB files - **highlight**: Code syntax highlighting implementation - **material3**: Material color utility extensions
EliasOulkadiAGENTS.mdEliasOulkadi/shokunin/.pack/AGENTS.md
HIPAA, DMCA. - **whendone-plus**: Notifications when long commands finish. ### Documents - **kami**: Professional PDF generation with parchment design system. - **portfolio-auto**: Automatic sync of GitHub repos to portfolio. - **kagen**: Render PDFs
liangdabiaoAGENTS.mdliangdabiao/exa-research-mcp-skill/AGENTS.md
progress tracker for long-running research. - `skills/renewable-market-research/references/`: detailed file-mode research, data-model, and PDF pipeline guidance. ## Editing Rules - Prefer updating `skills/` for cross-agent compatibility. - Do not add Claude
GoogleAGENTS.mdgoogle-gemini/gemini-managed-agents-templates/document-processor/AGENTS.md
files** (e.g. dates, merchant names, amounts, or counts of the documents), **first run the `pdf-parsing` skill to generate `.agents/workspace/parsed_invoices.json`** if it does not already exist. Then, write Python code
johnfkoo951AGENTS.mdjohnfkoo951/cmds-llm-wiki/AGENTS.md
Schema and harness document for the CMDS LLM Wiki vault. Defines the 3-layer architecture (Raw Sources / Wiki / Schema), ingest-query-lint operations, file conventions, and frontmatter standards. This is the single source of truth for LLM behavior in this vault.
ThreeMoonsLabAGENTS.mdThreeMoonsLab/agents-shipgate/AGENTS.md
# Agents Shipgate · Agent Instructions Authoritative instructions for AI coding agents (Claude Code, Codex, Cursor
LangChainAGENTS.mdlangchain-ai/paid-media-agent/AGENTS.md
renderer tokens aligned in the same change. Reuse those tokens in HTML, SVG, and PDF; do not introduce a separate palette or font in individual reports. Preserve the accepted appearance
gutomecAGENTS.mdgutomec/nirvana-os-engine/AGENTS.md
invoke the harness skill When the user asks for **any concrete artifact** — book, video, PDF, post, copy, design, illustration, brand, code, page, app, report, analysis, research, dataset, audit, anything — invoke
gutomecAGENTS.mdgutomec/nirvana-os-engine/skills/_shared/templates/AGENTS.md
invoke the harness skill When the user asks for **any concrete artifact** — book, video, PDF, post, copy, design, illustration, brand, code, page, app, report, analysis, research, dataset, audit, anything — invoke
p3botAGENTS.mdp3bot/snag/AGENTS.md
visible browser modes - Handle authenticated sessions gracefully - Multiple output formats: Markdown (default), HTML, Text, PDF, PNG - Multiple URL support: Process multiple URLs in a single command - Tab management: List, select
mgiffordAGENTS.mdmgifford/ACCESSIBILITY.md/AGENTS.md
Pages, editors, and export pipelines as separate surfaces. - Inspect final SVG, raster, print, and PDF output as applicable. ### Generated assets Record source revision, renderer or generator version, configuration, security profile
BENZEMA216AGENTS.mdBENZEMA216/ai-ecommerce-agent-skills/workspace-config/AGENTS.md
WITHIN 产品图：`/root/.openclaw/design-assets/seek-within/photos-inbound/` - SEEK WITHIN 参考图：`/root/.openclaw/design-assets/seek-within/references/` - SEEK WITHIN 原始拍摄：`/root/.openclaw/design-assets/seek-within/photos/` - SEEK WITHIN catalog PDF：`/root/.openclaw/design-assets/seek-within/catalogs/` - 资产索引：`/root/.openclaw/design-assets/seek-within/asset-catalog.json` ## Session 任务交接协议 **触发条件：** 任何时候任务进行到一半但还没完成（比如被打断、切换话题、或感觉 context 快满了） **必须执行：** 1. 把未完成任务写入
dobralabsAGENTS.mddobralabs/ccos-ratos/AGENTS.md
raiz não carrega a subpasta sozinho: é preciso ir buscar. - **Chegou material bruto** (transcrição, PDF, email exportado): vai pra pasta do projeto certo, e o essencial é destilado no `contexto.md
bybren-llcAGENTS.mdbybren-llc/story-systems-template/AGENTS.md
Agent (understands intent) ↓ WTFB Agents (craft expertise + authority) ↓ @wtfb/cli via Bash (execution) ↓ Output (PDF, FDX, HTML, validated files) ``` ### Example Interactions | Writer Says | Agent Runs | |-------------|------------| | "Export my script to PDF" | `wtfb
An open format for instructions to coding agents, read by Codex, Cursor and others. Think of it as a README written for agents.
At the repository root, with more specific files in subdirectories. Agents read the one closest to the file they're editing.
Setup and test commands, code style, and the rules a new contributor would need to know.
Claude Code reads CLAUDE.md. A one-line CLAUDE.md that points at AGENTS.md covers both.
