> ## Documentation Index > Fetch the complete documentation index at: https://docs.declaw.ai/llms.txt > Use this file to discover all available pages before exploring further. # Cookbook > Runnable examples for every Declaw feature, organized by category. Browse 49 end-to-end examples. Every example has a `main.py` (and `main.ts` where noted) you can run against your own Declaw deployment. ## Getting Started Create a sandbox, run a command, print output. Python + TypeScript. Create, inspect, extend timeout, kill. Both SDKs. Prove two sandboxes have fully isolated filesystems and processes. ## Built-in Templates Run `git`, `curl`, and `jq` end-to-end inside the minimal sandbox. Pipe a CSV through `pandas` and read JSON results back. Compile a `.ts` file with `tsc` and run it under Node 20. Verify the major LLM-framework SDKs import inside the agent sandbox. Fintech: four-agent CrewAI KYC pipeline with PII + injection defense. Health-tech: LangGraph workflow with PHI redact + rehydrate around a real GPT-4.1 call. ## Commands Run commands with env vars, working directories, and error handling. Stream command output in real-time via SSE callbacks. Start, list, and kill background processes by PID. Run shell, Python scripts, and pipelines in a single sandbox. ## Filesystem Write, read, list, rename, batch-write, and remove files. Both SDKs. Upload a CSV, run an analysis script, read back JSON results. Generate files in a sandbox and iterate over them on the host. ## Network Policies Block all outbound traffic. Compare against an open sandbox. Allow only specific domains; block everything else. Deny-all networking prevents data from leaving the sandbox. Block cloud metadata service (169.254.169.254) to prevent SSRF. ## PII Protection Configure PII scanning and redaction on outbound HTTP traffic. Compare redact, block, and log-only PII action modes. Transparent PII deanonymization in API responses. PII rehydration with SSE streaming responses and chunk buffering. ## Security Features Configure prompt injection detection and blocking. Regex-based request and response body transformations. Enable and retrieve security audit logs from a sandbox. ## LLM Providers GPT-4o-mini generates code; Declaw executes it safely. Claude generates code; Declaw executes it in an isolated sandbox. Llama via Groq generates code; Declaw executes it. Google Gemini generates code; Declaw executes it. Ollama or vLLM local LLM with Declaw code execution. ## Framework Integrations LangGraph ReAct agent with a Declaw `@tool`. CrewAI agent with a Declaw code execution tool. AutoGen `CodeExecutor` backed by Declaw sandboxes. OpenAI Agents SDK with a Declaw function tool. Agno (Phidata) toolkit wrapping Declaw sandboxes. Haystack pipeline component for Declaw code execution. Mastra agent with a Declaw tool — TypeScript only. ## Agent-in-Sandbox Upload and run an autonomous agent script inside a sandbox. OpenAI-powered agent running inside a locked-down sandbox. Anthropic Claude agent executing inside a sandbox. CrewAI multi-agent workflow running sandboxed. Agent in sandbox with network access restrictions. Agent with PII + injection defense + audit + network policy. ## Real-World Patterns Run tests in a sandbox to simulate a CI pipeline. Upload dataset, run analysis, download results. Both SDKs. Web scraping with a domain-restricted network policy. Run an HTTP API server inside a sandbox. Multi-agent pipeline where each agent runs in its own sandbox. Clone a repo, find a bug, apply a fix, run tests. ## Security Demos Prompt injection attack scenarios and how Declaw defends. Malicious package containment inside an isolated sandbox. PII scanning + deny-all prevents credential exfiltration. Side-by-side: unsecured execution vs. fully secured Declaw sandbox.