← 返回 Skills 市场

OpenClaw Boot Camp

Name: OpenClaw Boot Camp
Author: dank-varley

作者 dank-varley · GitHub ↗ · v1.2.1 · MIT-0

cross-platform ✓ 安全检测通过

157

总下载

当前安装

版本数

在 OpenClaw 中安装

/install bootcamp

功能描述

Auto-generates a CLI reference doc so your agent stops guessing OpenClaw commands and starts working. Discovers all commands, subcommands, and flags from you...

使用说明 (SKILL.md)

OpenClaw Boot Camp — Agent CLI Training

Every OpenClaw agent wastes tokens fumbling with CLI syntax. The --help command is broken on most subcommands. Flags get guessed wrong. Commands that don't exist get tried.

Boot Camp fixes that by discovering your actual CLI commands and generating a clean reference your agent can read.

Usage

bash bootcamp.sh

Follow the wizard. Your agent gets a cheat sheet at ~/.openclaw/workspace/notes/openclaw-cli-reference.md.

Modes

Local Only (Quickest) — Discovers commands from your CLI. No internet, no API calls, no token cost.
Local + Enrich (Recommended) — Local discovery + searches docs.openclaw.ai for your version. Still free.

What It Discovers

All top-level commands with descriptions
Every subcommand per command
Every flag per command
Which commands support --json
Doc links for key commands (Enrich mode)
Tips and gotchas baked in

Non-Interactive Mode

bash bootcamp.sh --yes --enrich

Version Management

Detects existing reference docs and offers to overwrite or keep both versions side-by-side.

Requirements

OpenClaw installed and in PATH
Bash 4+
No other dependencies

安全使用建议

This skill appears to do what it says: it runs your installed openclaw binary, parses the completion/help output, and writes a generated markdown reference to ~/.openclaw/workspace/notes. Before running: (1) inspect the bundled bootcamp.sh yourself (you already have it) to confirm there are no unexpected network calls in parts not shown; (2) be aware the script will call openclaw commands and may generate a completions file if missing (it does not overwrite an existing completion file); (3) only enable or allow the 'enrich' mode if you trust your agent's web access, since enrichment implies the agent may query docs.openclaw.ai or perform external searches; (4) back up any existing reference docs you care about. If you want extra caution, run the script in a contained environment or with verbose/tracing enabled to observe what openclaw subcommands it invokes.

功能分析

Type: OpenClaw Skill Name: bootcamp Version: 1.2.1 The 'bootcamp' skill is a utility designed to generate a CLI reference document for the OpenClaw agent. The core logic in `bootcamp.sh` scans local bash completion files and help outputs to map commands and flags, optionally enriching this data by fetching descriptions from official documentation at docs.openclaw.ai. The script uses standard text processing (grep, sed, and an inline Python parser) to create a markdown file in the agent's workspace. No indicators of data exfiltration, malicious execution, or prompt injection were found.

能力评估

✓ Purpose & Capability

Name/description claim to discover CLI commands and generate a reference; the included bootcamp.sh actually reads completion/help output, parses commands/flags, and writes a markdown file to the OpenClaw workspace. Required resources (openclaw in PATH, Bash 4+, read/write under ~/.openclaw) match the stated purpose.

ℹ Instruction Scope

The runtime instructions and script operate locally: they call openclaw --version, openclaw --help, and openclaw completion --shell bash, parse outputs, and write the generated reference to ~/.openclaw/workspace/notes. The README/SKILL.md advertise an optional 'Local + Enrich' agent-assisted phase that may consult docs.openclaw.ai; the supplied script (as shown) does not contain explicit network calls but implies the agent may perform web searches during enrichment — so expect external queries only if you enable enrichment or your agent has web access.

✓ Install Mechanism

No install spec; the skill is instruction-only with a bundled shell script. Nothing is downloaded or extracted from remote URLs during installation. Execution is local bash script usage.

✓ Credentials

The skill requires no environment variables, no secrets, and no unrelated credentials. It reads/writes under the OpenClaw state directory (~/.openclaw) which is proportional to generating an OpenClaw CLI reference.

✓ Persistence & Privilege

always is false and the skill does not request permanent platform privileges. It writes its output into the OpenClaw workspace and may create a completion file if none exists — this write behavior is consistent with the stated goal. The script does not modify other skills or global agent settings.

如何使用

确保已安装 OpenClaw（本地或 Docker 部署）
在对话框中输入安装命令：/install bootcamp
安装完成后，直接呼叫该 Skill 的名称或使用 /bootcamp 触发
根据 Skill 的参数说明提供必要输入，即可获得结构化输出

版本历史

v1.2.1

Patch: enrichment phase now filters mid-sentence HTML fragments and incomplete-thought captures (no more comma/colon-led noise lines). Added curl --connect-timeout 5 / --max-time 15 so the wizard can't hang on a flaky doc server.

v1.2.0

Enrich now dynamically discovers ALL commands with flags (21 up from 9). Fetches full inline flag details from docs, not just links. 30 flags per page cap.

v1.1.0

Enrich mode now fetches inline flag details from docs instead of just links. Progress bar added. Wizard text polished. Next step guidance added.

v1.0.0

Initial release: CLI reference generator with wizard UI, local discovery + docs enrichment

元数据

Slug bootcamp

版本 1.2.1

许可证 MIT-0

累计安装 0

当前安装数 0

历史版本数 4

常见问题