docker-sandbox
Create, manage, and execute agent tools (claude, codex) inside Docker sandboxes for isolated code execution. Use when running agent loops, spawning tool subprocesses, or any task requiring process isolation. Triggers on "sandbox", "isolated execution", "docker sandbox", "safe agent execution", or when working on agent loop infrastructure.
92
88%
Does it follow best practices?
Impact
100%
3.22xAverage score across 3 eval scenarios
Risky
Do not use without reviewing
Quality
Discovery
100%Based on the skill's description, can an agent find and select it at the right time? Clear, specific descriptions lead to better discovery.
This is a strong skill description that clearly articulates specific capabilities (creating and managing agent tools in Docker sandboxes), provides explicit 'Use when' guidance with multiple trigger scenarios, and includes a list of natural trigger terms. The description is concise, uses third-person voice, and occupies a distinct niche that minimizes conflict risk with other skills.
| Dimension | Reasoning | Score |
|---|---|---|
Specificity | Lists multiple specific concrete actions: 'Create, manage, and execute agent tools (claude, codex) inside Docker sandboxes for isolated code execution.' This names specific tools, the container technology, and the purpose clearly. | 3 / 3 |
Completeness | Clearly answers both 'what' (create, manage, execute agent tools in Docker sandboxes) and 'when' (explicit 'Use when' clause covering agent loops, spawning tool subprocesses, process isolation tasks, plus explicit trigger terms). | 3 / 3 |
Trigger Term Quality | Includes strong natural trigger terms: 'sandbox', 'isolated execution', 'docker sandbox', 'safe agent execution', 'agent loop infrastructure', 'agent loops', 'tool subprocesses', 'process isolation'. These cover a good range of terms a user would naturally use. | 3 / 3 |
Distinctiveness Conflict Risk | Highly distinctive niche combining Docker sandboxes with agent tool execution. The specific mention of 'claude, codex' tools, 'Docker sandboxes', and 'agent loop infrastructure' makes it very unlikely to conflict with generic Docker or generic agent skills. | 3 / 3 |
Total | 12 / 12 Passed |
Implementation
77%Reviews the quality of instructions and guidance provided to agents. Good implementation is clear, handles edge cases, and produces reliable results.
This is a strong, actionable skill with excellent workflow clarity and concrete executable commands throughout. Its main weakness is length—the inline TypeScript implementation signatures and detailed auth setup could be split into referenced files to improve progressive disclosure and conciseness. The quick reference section at the top is well-designed for fast access.
Suggestions
Move the 'Implementation in utils.ts' section to a separate referenced file (e.g., IMPLEMENTATION.md) since it describes code to write rather than providing immediately executable guidance.
Move the detailed 'Auth Setup (One-Time)' section to a separate AUTH.md file, keeping only a brief summary and link in the main skill.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | The skill is fairly comprehensive and mostly efficient, but includes some sections that could be tightened—like the Implementation in utils.ts section with TypeScript signatures that describe what to build rather than providing ready-to-use code, and the timing tables which add bulk. Some explanatory text (e.g., 'Uses existing Claude Max and ChatGPT Pro subscriptions — no API key billing') is unnecessary context. | 2 / 3 |
Actionability | The skill provides concrete, executable bash commands throughout—sandbox creation, auth setup, exec commands, network control, template saving, and troubleshooting diagnostics. Commands are copy-paste ready with realistic flags and arguments. The TypeScript section is less executable but the core operational guidance is fully actionable. | 3 / 3 |
Workflow Clarity | The agent loop integration section clearly sequences the pre-warm pattern (create → inject auth → exec per story → destroy), with explicit lifecycle management. The auth setup has clear sequential steps. The fallback to host mode provides error recovery. The troubleshooting section serves as validation checkpoints for common failure modes. | 3 / 3 |
Progressive Disclosure | The content is well-structured with clear section headers and a quick reference at the top, but it's quite long (~180 lines of substantive content) with the TypeScript implementation details, auth setup, and troubleshooting all inline. The auth setup and implementation sections could be split into separate referenced files. The ADR link is a good example of progressive disclosure, but more content could be offloaded. | 2 / 3 |
Total | 10 / 12 Passed |
Validation
90%Checks the skill against the spec for correct structure and formatting. All validation checks must pass before discovery and implementation can be scored.
Validation — 10 / 11 Passed
Validation for skill structure
| Criteria | Description | Result |
|---|---|---|
frontmatter_unknown_keys | Unknown frontmatter key(s) found; consider removing or moving to metadata | Warning |
Total | 10 / 11 Passed | |
- Repository
- joelhooks/joelclaw
- Commit
ce9ca8e
- Reviewed
Table of Contents
Is this your skill?
If you maintain this skill, you can claim it as your own. Once claimed, you can manage eval scenarios, bundle related skills, attach documentation or rules, and ensure cross-agent compatibility.