Skip to content
Back to search

agent-sandbox-skill

An agent skill for managing isolated execution environments

Safety scan

A static text scan, not a guarantee. Select the badge to see what was scanned and every finding.

Checked today
Badge for your README
findsafeskills safety rating

Markdown

[![findsafeskills safety rating](/api/listings/disler--agent-sandbox-skill/badge.svg)](/listings/disler--agent-sandbox-skill)

HTML

<a href="/listings/disler--agent-sandbox-skill"><img src="/api/listings/disler--agent-sandbox-skill/badge.svg" alt="findsafeskills safety rating" /></a>

Install

i

Install instructions from the README

As written by the repo's author. findsafeskills has not verified these steps.

Prerequisites

* Python >= 3.12 * `uv` package manager (recommended) * E2B Account & API Key

Installation

1. **Clone the repository** (if not already done). 2. **Configure Environment**: Create a `.env` file in the project root: ```bash cp .env.sample .env ``` Add your E2B API key: ```env E2B_API_KEY=sbx_... ``` *(Get your key from [E2B Dashboard](https://e2b.dev/dashboard/keys))*

3. **Open your Agentic Coding Tool**: Open your Agentic Coding Tool and start prompting!

```bash claude or gemini or codex ```

Then for one of simple tasks run:

```bash \sandbox <prompt> ```

For complex tasks run ```bash \plan-build-host-test <prompt> <workflow_id> ```

What was scanned

The files the safety scan read, as written by the repository's author and shown as plain text. Lines with a finding are marked.

README.md · README · 223 lines

View this file on GitHub (read at the repository's current version: the scan did not record which commit it read)

  1. # Agent Sandbox Skill
  2. An agent skill for managing isolated execution environments using [E2B Sandboxes](https://e2b.dev/).
  3. This skill enables AI agents (Gemini CLI, Claude Code, Codex CLI) to safely execute code, build full-stack applications, and perform arbitrary engineering tasks in a secure, isolated sandbox.
  4. Watch the [Gemini 3 Demo](https://youtu.be/V5IhsHEHXOg) or the newer [Claude Opus 4.5](https://youtu.be/3kgx0YxCriM) Demo to understand what this codebase/skill can do for your agentic coding.
  5. <img src="images/sandboxes.png" alt="Agent Sandboxes" width="800">
  6. ## Why Use Agent Sandboxes? The **Value Proposition**
  7. >
  8. > Agent Sandboxes unlock 3 key capabilities for your agentic engineering:
  9. - **Isolation**: Each agent fork runs in a fully isolated, gated E2B sandbox, this means no matter what your agent does, it's secure and safe from your local filesystem and production environment.
  10. - **Scale**: You can run as many agent forks as you want, each fork is independent and has its own sandbox. This is a very literal way to scale your compute to scale your impact.
  11. - **Agency**: Your agents have full control over the sandbox environment, they can install packages, modify files, run commands, etc. This means they can handle more of the engineering process for you.
  12. ## 🚀 Features
  13. * **Isolated Execution**: Run untrusted code, tests, and binaries safely.
  14. * **Full-Stack Development**: Scaffold, build, and host Vue + FastAPI + SQLite apps.
  15. * **Browser Automation**: Built-in Playwright integration for visual validation.
  16. * **Agent-First Design**: Optimized for CLI agents with structured prompts and robust error handling.
  17. * **Persistent Context**: Tools to manage sandbox lifecycles across agent turns.
  18. ## 🛠️ Setup
  19. ### Prerequisites
  20. * Python >= 3.12
  21. * `uv` package manager (recommended)
  22. * E2B Account & API Key
  23. ### Installation
  24. 1. **Clone the repository** (if not already done).
  25. 2. **Configure Environment**:
  26. Create a `.env` file in the project root:
  27. ```bash
  28. cp .env.sample .env
  29. ```
  30. Add your E2B API key:
  31. ```env
  32. E2B_API_KEY=sbx_...
  33. ```
  34. *(Get your key from [E2B Dashboard](https://e2b.dev/dashboard/keys))*
  35. 3. **Open your Agentic Coding Tool**:
  36. Open your Agentic Coding Tool and start prompting!
  37. ```bash
  38. claude
  39. or
  40. gemini
  41. or
  42. codex
  43. ```
  44. Then for one of simple tasks run:
  45. ```bash
  46. \sandbox <prompt>
  47. ```
  48. For complex tasks run
  49. ```bash
  50. \plan-build-host-test <prompt> <workflow_id>
  51. ```
  52. ## Starter Prompts
  53. The `prompts/{claude,gemini,codex}/<difficulty>_<task>.md` contain full stack application prompts you can
  54. Try these prompts out with a SOTA Model for the best results.
  55. Run these prompts in your Agentic Coding Tool.
  56. Start with the Very Easy Prompts to get a feel for the tool. Keep in mind token usage increases as the complexity of the prompt increases.
  57. ### Very Easy Prompts
  58. ```bash
  59. \agent-sandboxes:plan-build-host-test "$(cat prompts/full_stack/codex/very_easy_guestbook.md)" "very_easy_guestbook"
  60. \agent-sandboxes:plan-build-host-test "$(cat prompts/full_stack/codex/very_easy_url_shortener.md)" "very_easy_url_shortener"
  61. \agent-sandboxes:plan-build-host-test "$(cat prompts/full_stack/gemini/very_easy_counter.md)" "very_easy_counter"
  62. \agent-sandboxes:plan-build-host-test "$(cat prompts/full_stack/gemini/very_easy_poll_maker.md)" "very_easy_poll_maker"
  63. \agent-sandboxes:plan-build-host-test "$(cat prompts/full_stack/sonnet/very_easy_calculator.md)" "very_easy_calculator"
  64. \agent-sandboxes:plan-build-host-test "$(cat prompts/full_stack/sonnet/very_easy_counter.md)" "very_easy_counter"
  65. \agent-sandboxes:plan-build-host-test "$(cat prompts/full_stack/sonnet/very_easy_greeter.md)" "very_easy_greeter"
  66. ```
  67. ### Easy Prompts
  68. ```bash
  69. \agent-sandboxes:plan-build-host-test "$(cat prompts/full_stack/gemini/easy_api_mock_studio.md)" "easy_api_mock_studio"
  70. \agent-sandboxes:plan-build-host-test "$(cat prompts/full_stack/gemini/easy_code_snippet_manager.md)" "easy_code_snippet_manager"
  71. \agent-sandboxes:plan-build-host-test "$(cat prompts/full_stack/gemini/easy_cron_heartbeat_monitor.md)" "easy_cron_heartbeat_monitor"
  72. \agent-sandboxes:plan-build-host-test "$(cat prompts/full_stack/gemini/easy_markdown_knowledge_base.md)" "easy_markdown_knowledge_base"
  73. \agent-sandboxes:plan-build-host-test "$(cat prompts/full_stack/gemini/easy_offline_task_board.md)" "easy_offline_task_board"
  74. \agent-sandboxes:plan-build-host-test "$(cat prompts/full_stack/gemini/easy_schema_visualizer.md)" "easy_schema_visualizer"
  75. \agent-sandboxes:plan-build-host-test "$(cat prompts/full_stack/sonnet/easy_nano_banana_simple.md)" "easy_nano_banana_simple"
  76. \agent-sandboxes:plan-build-host-test "$(cat prompts/full_stack/sonnet/easy_notes_app.md)" "easy_notes_app"
  77. \agent-sandboxes:plan-build-host-test "$(cat prompts/full_stack/sonnet/easy_sqlite_crud.md)" "easy_sqlite_crud"
  78. \agent-sandboxes:plan-build-host-test "$(cat prompts/full_stack/sonnet/easy_todo_list.md)" "easy_todo_list"
  79. \agent-sandboxes:plan-build-host-test "$(cat prompts/full_stack/sonnet/easy_chart_sketch.md)" "easy_chart_sketch"
  80. \agent-sandboxes:plan-build-host-test "$(cat prompts/full_stack/sonnet/easy_decision_matrix.md)" "easy_decision_matrix"
  81. \agent-sandboxes:plan-build-host-test "$(cat prompts/full_stack/sonnet/easy_design_system_palette.md)" "easy_design_system_palette"
  82. ```
  83. ### Medium Prompts
  84. ```bash
  85. \agent-sandboxes:plan-build-host-test "$(cat prompts/full_stack/codex/medium_knowledge_base_curator.md)" "medium_knowledge_base_curator"
  86. \agent-sandboxes:plan-build-host-test "$(cat prompts/full_stack/gemini/medium_devops_dashboard.md)" "medium_devops_dashboard"
  87. \agent-sandboxes:plan-build-host-test "$(cat prompts/full_stack/gemini/medium_investment_tracker.md)" "medium_investment_tracker"
  88. \agent-sandboxes:plan-build-host-test "$(cat prompts/full_stack/gemini/medium_log_analysis_tool.md)" "medium_log_analysis_tool"
  89. \agent-sandboxes:plan-build-host-test "$(cat prompts/full_stack/gemini/medium_planning_poker.md)" "medium_planning_poker"
  90. \agent-sandboxes:plan-build-host-test "$(cat prompts/full_stack/opus45/medium_elevenlabs_live_transcription_app.md)" "medium_elevenlabs_live_transcription_app"
  91. \agent-sandboxes:plan-build-host-test "$(cat prompts/full_stack/sonnet/medium_file_explorer.md)" "medium_file_explorer"
  92. \agent-sandboxes:plan-build-host-test "$(cat prompts/full_stack/sonnet/medium_habit_tracker.md)" "medium_habit_tracker"
  93. \agent-sandboxes:plan-build-host-test "$(cat prompts/full_stack/sonnet/medium_nano_banana_generator.md)" "medium_nano_banana_generator"
  94. \agent-sandboxes:plan-build-host-test "$(cat prompts/full_stack/sonnet/medium_personal_finance.md)" "medium_personal_finance"
  95. \agent-sandboxes:plan-build-host-test "$(cat prompts/full_stack/sonnet/medium_recipe_planner.md)" "medium_recipe_planner"
  96. \agent-sandboxes:plan-build-host-test "$(cat prompts/full_stack/sonnet/medium_chart_sketch_pro.md)" "medium_chart_sketch_pro"
  97. ```
  98. ### Hard Prompts
  99. ```bash
  100. \agent-sandboxes:plan-build-host-test "$(cat prompts/full_stack/codex/hard_incident_response_notebook.md)" "hard_incident_response_notebook"
  101. \agent-sandboxes:plan-build-host-test "$(cat prompts/full_stack/codex/hard_learning_cohort_orchestrator.md)" "hard_learning_cohort_orchestrator"
  102. \agent-sandboxes:plan-build-host-test "$(cat prompts/full_stack/codex/hard_supplier_quality_portal.md)" "hard_supplier_quality_portal"
  103. \agent-sandboxes:plan-build-host-test "$(cat prompts/full_stack/codex/hard_ui_design_review_platform.md)" "hard_ui_design_review_platform"
  104. \agent-sandboxes:plan-build-host-test "$(cat prompts/full_stack/sonnet/hard_api_testing.md)" "hard_api_testing"
  105. \agent-sandboxes:plan-build-host-test "$(cat prompts/full_stack/sonnet/hard_code_snippets.md)" "hard_code_snippets"
  106. \agent-sandboxes:plan-build-host-test "$(cat prompts/full_stack/sonnet/hard_content_workflow.md)" "hard_content_workflow"
  107. \agent-sandboxes:plan-build-host-test "$(cat prompts/full_stack/sonnet/hard_freelancer_manager.md)" "hard_freelancer_manager"
  108. \agent-sandboxes:plan-build-host-test "$(cat prompts/full_stack/sonnet/hard_nano_banana_studio.md)" "hard_nano_banana_studio"
  109. \agent-sandboxes:plan-build-host-test "$(cat prompts/full_stack/sonnet/hard_time_tracking.md)" "hard_time_tracking"
  110. ```
  111. ### Very Hard Prompts
  112. ```bash
  113. \agent-sandboxes:plan-build-host-test "$(cat prompts/full_stack/codex/very_hard_analytics_ops_monitoring_hub.md)" "very_hard_analytics_ops_monitoring_hub"
  114. \agent-sandboxes:plan-build-host-test "$(cat prompts/full_stack/codex/very_hard_iot_fleet_maintenance_console.md)" "very_hard_iot_fleet_maintenance_console"
  115. \agent-sandboxes:plan-build-host-test "$(cat prompts/full_stack/codex/very_hard_micro_betting_odds_lab.md)" "very_hard_micro_betting_odds_lab"
  116. \agent-sandboxes:plan-build-host-test "$(cat prompts/full_stack/codex/very_hard_personal_investing_allocator.md)" "very_hard_personal_investing_allocator"
  117. \agent-sandboxes:plan-build-host-test "$(cat prompts/full_stack/codex/very_hard_scriptwriter_automation_workbench.md)" "very_hard_scriptwriter_automation_workbench"
  118. \agent-sandboxes:plan-build-host-test "$(cat prompts/full_stack/sonnet/very_hard_knowledge_base.md)" "very_hard_knowledge_base"
  119. \agent-sandboxes:plan-build-host-test "$(cat prompts/full_stack/sonnet/very_hard_system_monitor.md)" "very_hard_system_monitor"
  120. ```
  121. ## 3 Browser Testing Prompts
  122. > after you host your app via public url, you can test it with the browser testing prompts.
  123. ```bash
  124. # Claude Code only - requires subagent support (parallel=true, headed=true)
  125. /generic-browser-test https://www.anthropic.com/news/claude-opus-4-5 prompts/browser-workflows/opus-4-5-release.md true true
  126. # Sandbox app browser tests (replace <URL> with actual sandbox URL)
  127. /generic-browser-test <URL> prompts/browser-workflows/easy_chart_sketch_browser_test.md true true
  128. /generic-browser-test <URL> prompts/browser-workflows/easy_decision_matrix_browser_test.md true true
  129. /generic-browser-test <URL> prompts/browser-workflows/easy_design_system_palette_browser_test.md true true
  130. /generic-browser-test <URL> prompts/browser-workflows/medium_chart_sketch_pro_browser_test.md true true
  131. # For other agentic coding tools, make sure subagents is false
  132. \generic-browser-test <URL> prompts/browser-workflows/easy_chart_sketch_browser_test.md false true
  133. ```
  134. ## 🤖 "Reprogrammed" BACKSLASH Commands
  135. This project uses "reprogrammed" backslash commands to trigger agent workflows.
  136. See [CLAUDE.md](CLAUDE.md), [AGENTS.md](AGENTS.md), and [GEMINI.md](GEMINI.md) for more details.
  137. | Command | Description |
  138. | :------------------------------------------------------------------------------ | :------------------------------------------------------------------------------------------- |
  139. | `\sandbox <prompt>` | Small general sandbox operations. An adhoc prompt with minimal compute usage. |
  140. | `\agent-sandboxes:plan-full-stack <prompt>` | Generates a detailed implementation plan with Browser UI Testing workflows. |
  141. | `\agent-sandboxes:build <plan_path>` | Executes a build plan within a sandbox. |
  142. | `\agent-sandboxes:host <sandbox_id> <port>` | Exposes a port and returns a public URL. |
  143. | `\agent-sandboxes:test <sandbox_id> <url> <plan_path> <workflow_id>` | Runs validation tests against the hosted app including browser UI testing. |
  144. | `\agent-sandboxes:browser-testing <sandbox_id> <url> <plan_path> <workflow_id>` | Executes browser UI testing workflows from a plan using parallel subagents (headed default). |
  145. | `\agent-sandboxes:plan-build-host-test <prompt> <workflow_id>` | **Agentic-Workflow**: Orchestrates a full lifecycle: Plan → Build → Host → Test. |
  146. ## 💻 CLI Usage (Manual)
  147. You can also run the CLI manually for debugging or inspection:
  148. ```bash
  149. # From .claude/skills/agent-sandboxes/sandbox_cli/
  150. uv run sbx --help
  151. ```
  152. ### Common Commands
  153. * **Init**: `uv run sbx init --timeout 1800`
  154. * **Execute**: `uv run sbx exec <sandbox_id> "echo hello"`
  155. * **Files**: `uv run sbx files ls <sandbox_id> /home/user`
  156. * **Browser**: `uv run sbx browser start`
  157. ## 🏗️ Architecture
  158. * **CLI**: Python-based (`click`, `e2b`, `rich`) located in `.claude/skills/agent-sandboxes/sandbox_cli/`.
  159. * **Prompts**: Markdown-based prompt templates in `.claude/skills/agent-sandboxes/prompts/`.
  160. * **Skill Definition**: `SKILL.md` defines the capabilities exposed to the agent.
  161. ## 📄 Documentation
  162. * [Skill Definition & Guide](.claude/skills/agent-sandboxes/SKILL.md)
  163. * [CLI README](.claude/skills/agent-sandboxes/sandbox_cli/README.md)
  164. * [E2B Documentation](https://e2b.dev/docs)
  165. ## Master **Agentic Coding**
  166. > Prepare for the future of software engineering
  167. Learn tactical agentic coding patterns with [Tactical Agentic Coding](https://agenticengineer.com/tactical-agentic-coding?y=agsbxsk)
  168. Follow the [IndyDevDan YouTube channel](https://www.youtube.com/@indydevdan) to improve your agentic coding advantage.

Ratings and comments

Sign in to rate this skill.

Sign in

Free to use. If it’s helped, pass it on.