據 AlphaMoat 統計,截至 2026-09,PaperPod 累計下載 2,564 次、獲 2 個星標,在全部 5,078 個 AI 技能中排名第 4,045 位,在「AI 智慧體」分類(969 個)中排名第 815 位。
| 平台 | 累計安裝 | 累計下載 | 星標 | 評論 |
|---|---|---|---|---|
| ClawHub ↗ | 74 | 2564 | 2 | - |
| SkillHub ↗ | 134 | 2504 | 2 | — |
各平台的計數口徑不同,統計時間不一致。
該服務可執行後臺程序並生成公開預覽地址;令牌、憑據和私人資料可能因命令執行或持久儲存而暴露。
資料釋出於 2026-10-29
以下是根據技能文件生成的示例,不代表實際執行結果。
2026-09 起統計方法調整:同名技能按作者分別統計,並覆蓋全部技能。圖中虛線為調整前的月份,與之後的資料不直接比較,也不計算跨越調整的增長。
| 月份 | 累計下載 | 環比 | 星標 |
|---|---|---|---|
| 2026-09 | 2564 | — | 2 |
| 2026-08 | 2431 | +6.9% | 2 |
| 2026-07 | 2275 | +14.1% | 2 |
| 2026-06 | 1993 | +8.4% | 2 |
| 2026-05 | 1838 | — | 2 |
| 2026-03 | 1400 | — | 2 |
以下是作者釋出的原始檔案,按純文字顯示,本站未做修改或稽核;其中的連結不可點選,命令請核實後再執行。抓取於 2026-10-23。
--- name: paperpod description: Isolated agent runtime for code execution, live preview URLs, browser automation, 50+ tools (ffmpeg, sqlite, pandoc, imagemagick), LLM inference, and persistent memory — all via CLI or HTTP, no SDK or API keys required. metadata: author: PaperPod version: "2.0.3" homepage: https://paperpod.dev --- # PaperPod
Isolated, agent-native sandboxes for code execution, live preview URLs, browser automation, 50+ tools (ffmpeg, sqlite, pandoc, imagemagick), LLM inference, and persistent memory — all via CLI or HTTP, no SDK or API keys required. Billed per second of compute usage and topup via stripe or x402.
## Quick Start
```bash
curl -X POST https://paperpod.dev/login -d '{"email":"[email protected]"}' # Verify and get token
npm install -g @paperpod/cli
ppod login <token> && ppod help
```
## Authentication
**Step 1: Get a token**
```bash
curl -X POST https://paperpod.dev/login -d '{"email":"[email protected]"}'
# Check email → click magic link → copy token (pp_sess_...)
```
**Step 2: Use the token** (pick one method)
| Method | How | Best for |
|--------|-----|----------|
| **CLI login** | `ppod login pp_sess_...` | Everyday, Interactive use |
| **Env var** | `export PAPERPOD_TOKEN=pp_sess_...` | Scripts, CI/CD |
| **Per-request** | `-H "Authorization: Bearer pp_sess_..."` | HTTP one-shots |
Tokens expire in **15 days**. On `EXPIRED_TOKEN` error, re-authenticate via `POST /login`.
---
## CLI (Recommended)
The CLI is the easiest way to use PaperPod. It handles streaming, sessions, and reconnection automatically.
### CLI Commands
| Category | Command | Description |
|----------|---------|-------------|
| **Sandbox** | `ppod exec <cmd>` | Run shell command |
| | `ppod write <path> [file]` | Write file (stdin if no file) |
| | `ppod read <path>` | Read file |
| | `ppod ls <path>` | List directory |
| **Processes** | `ppod start <cmd>` | Start background process |
| | `ppod ps` | List processes |
| | `ppod kill <id>` | Stop process |
| **Ports** | `ppod expose <port>` | Get public URL (-q for URL only) |
| **Browser** | `ppod browser:screenshot <url>` | Capture webpage |
| | `ppod browser:pdf <url>` | Generate PDF |
| | `ppod browser:scrape <url> [sel]` | Scrape elements (default: body) |
| | `ppod browser:markdown <url>` | Extract markdown |
| | `ppod browser:content <url>` | Get rendered HTML |
| | `ppod browser:test <url> '<json>'` | Run Playwright tests |
| | `ppod browser:acquire` | Acquire reusable session |
| | `ppod browser:connect <id>` | Connect to existing session |
| | `ppod browser:sessions` | List active sessions |
| | `ppod browser:limits` | Check browser limits |
| **AI** | `ppod ai <prompt>` | Text generation |
| | `ppod ai:embed <text>` | Generate embeddings |
| | `ppod ai:image <prompt>` | Generate image |
| | `ppod ai:transcribe <audio>` | Transcribe audio |
| | `ppod ai:models` | List available AI models |
| **Code** | `ppod interpret <code>` | Rich output (charts) |
| **Memory** | `ppod mem:write <path>` | Persist data |
| | `ppod mem:read <path>` | Read persisted data |
| | `ppod mem:ls` | List memory files |
| | `ppod mem:rm <path>` | Delete from memory |
| | `ppod mem:usage` | Check quota |
| **Account** | `ppod balance` | Check credits |
| | `ppod status` | Connection info |
| | `ppod help` | Show all commands |
| | `ppod <cmd> --help` | Help for specific command |
**Update CLI:** `npm update -g @paperpod/cli`
### CLI Examples
```bash
# Execute code
ppod exec "python -c 'print(2+2)'"
ppod exec "npm init -y && npm install express"
# Start server + expose (--bind 0.0.0.0 required for public access)
ppod start "python -m http.server 8080 --bind 0.0.0.0"
ppod expose 8080 # → https://8080-{sandbox-id}-p8080_v1.paperpod.work (stable URL)
# Browser with tracing
ppod browser:screenshot https://example.com --trace debug.zip
# Persistent storage (survives sandbox reset)
echo '{"step":3}' | ppod mem:write state.json
# Built-in tools (50+ available: ffmpeg, sqlite3, pandoc, imagemagick, git, jq, ripgrep...)
ppod exec "ffmpeg -i input.mp4 -vf scale=640:480 output.mp4" # Video processing
ppod exec "sqlite3 data.db 'SELECT * FROM users'" # Database queries
ppod exec "convert image.png -resize 50% thumbnail.png" # Image manipulation
```
---
## HTTP Endpoints
Use HTTP for one-shot tasks or when CLI isn't available. Run `curl https://paperpod.dev/docs` or visit https://paperpod.dev/docs for full API reference.
### Quick Reference
| Endpoint | Purpose |
|----------|---------|
| `POST /execute` | Run code (python, javascript, shell) |
| `POST /execute/stream` | Stream output (SSE) |
| `POST /files/write` | Write file |
| `POST /files/read` | Read file |
| `POST /files/list` | List directory |
| `POST /process/start` | Start background process |
| `POST /process/list` | List processes |
| `POST /expose` | Get preview URL for port |
| `POST /memory/write` | Persist data |
| `POST /memory/read` | Read persisted data |
| `POST /browser/screenshot` | Capture screenshot |
| `POST /browser/pdf` | Generate PDF |
| `POST /browser/markdown` | Extract markdown |
| `POST /ai/generate` | Text generation |
| `POST /ai/embed` | Embeddings |
| `POST /ai/image` | Image generation |
| `GET /ai/models` | List models |
### HTTP Example
```bash
# Execute shell command
curl -X POST https://paperpod.dev/execute \
-H "Authorization: Bearer $PAPERPOD_TOKEN" \
-d '{"code": "ls -la", "language": "shell"}'
```
---
## Capabilities
| Category | What you can do |
|----------|-----------------|
| **Code Execution** | Python, JavaScript, shell commands |
| **Processes** | Background servers, long-running jobs |
| **Preview URLs** | Expose ports → `https://8080-{sandbox-id}-p8080_v1.paperpod.work` |
| **Agent Memory** | 10MB persistent storage (R2) |
| **Browser** | Screenshots, PDFs, scraping (Playwright) |
| **AI Models** | Text, embeddings, images, transcription |
| **Files** | Read/write, git, bulk operations |
### Pre-installed Tools (50+)
| Category | Tools |
|----------|-------|
| **Runtimes** | python, node, npm, bun, pip |
| **Version Control** | git, gh (GitHub CLI) |
| **HTTP & Networking** | curl, httpie, jq, dig, ss |
| **Search & Text** | ripgrep (rg), find, sed, awk, tree |
| **Media & Docs** | ffmpeg, imagemagick, pandoc |
| **Build & Data** | make, sqlite3, tar, gzip, zip, unzip |
## Key Notes
- **Sandboxes are isolated** — each user gets their own container with a full Linux environment; you can only affect your own ephemeral sandbox
- **Sandbox is ephemeral** — use Agent Memory (`/memory/*`) for persistence
- **Working directory is `/workspace`** — relative paths like `file.txt` resolve to `/workspace/file.txt`
- **Servers must bind to `0.0.0.0`** for public access
- **Ports 3000-3010 are reserved** — use 8080, 5000, 4000, etc.
- **Browser sessions** — Each command creates an ephemeral session. Use `browser:acquire` for multi-command session reuse, `--trace` to capture Playwright traces
## Billing
**$0.0001/sec** compute + browser, **$0.02/1K neurons** AI. New accounts get **$5 free** (~14 hours), no credit card required.
## Discovery
- `ppod help` — CLI command reference
- `GET https://paperpod.dev/` — API schema (JSON)
- `GET https://paperpod.dev/docs` — Full documentation
---
## Advanced: WebSocket (not recommended for normal workflows). For programmatic integrations or custom apps, connect via WebSocket. `GET https://paperpod.dev/docs` to learn more.
將開發過程中的經驗、錯誤與修正記錄為 Markdown 檔案,供日後重用並提升為工作區記憶。
用自然語言描述需求,在本機與遠端技能市場尋找技能並取得推薦。
給 AI agent 用的瀏覽器自動化命令列技能。
將使用者明確糾正與自我檢視整理成分層記憶,並依使用情況升級或封存。
在安裝第三方技能之前,按清單審查來源、許可權與危險模式的技能。
將代理記憶與跨技能狀態整理成具型別與約束的知識圖譜。
規劃 AI 代理主動跟進、工作記憶與回顧流程的工作區設定指南。
指導如何建立與更新有效 Claude 技能的寫作指南。
資料月份:2026-09 · 下載/星標/安裝數聚合自公開技能註冊庫(ClawHub、SkillHub),按月更新。口徑說明見資料方法。