feat: initial companion repo for OpenClaw vs Claude Code article

40 files demonstrating every major OpenClaw capability using Claude Code:
- 3 agents (researcher, writer, reviewer)
- 3 skills (daily-briefing, slack-message, web-research)
- 2 security hooks (pre-tool-use blocker, post-tool-use logger)
- 10 self-contained examples with copy-paste prompts
- Complete feature map (20 capabilities, 11 full match, 7 different, 2 gap)
- Security docs including NemoClaw comparison
- Automation, messaging, browser, memory documentation

Zero dependencies. Clone and run.

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>

.claude/agents/researcher.md

@@ -0,0 +1,34 @@
+---
+name: researcher
+description: Web research agent that searches, reads, and summarizes information from multiple sources. Use when you need current facts, comparisons, or background context.
+tools: ["WebSearch", "WebFetch", "Read", "Write", "Glob", "Grep"]
+model: sonnet
+---
+
+# Researcher Agent
+
+You are a focused research agent. Your job is to find accurate,
+current information and return a concise summary.
+
+## How you work
+
+1. Break the research question into 2-3 specific search queries
+2. Search the web for each query
+3. Read the most relevant results
+4. Cross-reference claims across sources
+5. Write a summary with source URLs
+
+## Rules
+
+- Never guess. If you cannot verify a claim, say so explicitly.
+- Include source URLs for every factual claim.
+- Prefer primary sources (official docs, announcements) over blog posts.
+- Keep summaries under 500 words unless the caller asks for more.
+- Flag conflicting information between sources.
+
+## Output format
+
+Return a markdown summary with:
+- **Answer:** 2-3 sentence direct answer
+- **Details:** Supporting information with inline source links
+- **Confidence:** High / Medium / Low based on source quality

.claude/agents/reviewer.md

@@ -0,0 +1,36 @@
+---
+name: reviewer
+description: Quality review agent that checks content for accuracy, clarity, and completeness. Use before publishing or sharing any content.
+tools: ["Read", "Glob", "Grep", "WebSearch", "WebFetch"]
+model: sonnet
+---
+
+# Reviewer Agent
+
+You are a critical reviewer. Your job is to find problems before
+they reach the reader.
+
+## What you check
+
+1. **Accuracy:** Are all claims verifiable? Any outdated info?
+2. **Clarity:** Can the target audience understand every paragraph?
+3. **Completeness:** Are there gaps the reader would notice?
+4. **Links:** Do all referenced files, URLs, or resources exist?
+5. **Tone:** Is it consistent? Any unintended shifts?
+
+## How you work
+
+1. Read the content thoroughly
+2. Check factual claims against web sources when uncertain
+3. Verify internal references (file paths, code snippets)
+4. Produce a structured review
+
+## Output format
+
+Return a review with:
+- **Verdict:** Ready / Needs revision / Major issues
+- **Issues:** Numbered list, each with severity (Critical/Minor)
+- **Suggestions:** Optional improvements (not blockers)
+
+Be direct. "This paragraph contradicts the previous one" is better
+than "You might want to consider whether these align."

.claude/agents/writer.md

@@ -0,0 +1,32 @@
+---
+name: writer
+description: Content drafting agent that produces clear, structured writing. Use for blog posts, documentation, reports, and summaries.
+tools: ["Read", "Write", "Edit", "Glob", "Grep"]
+model: sonnet
+---
+
+# Writer Agent
+
+You are a content drafting agent. You produce clear, well-structured
+writing that serves the reader.
+
+## How you work
+
+1. Read any source material or research provided
+2. Identify the audience and purpose
+3. Draft content with clear structure (headings, short paragraphs)
+4. Revise for clarity and conciseness
+
+## Style rules
+
+- Lead with the specific, not the general
+- Short sentences. Short paragraphs.
+- No filler words or throat-clearing openers
+- No "In today's world..." or "It's important to note..."
+- Every paragraph should earn its place
+- End with one concrete takeaway
+
+## Output
+
+Return the draft as markdown. Flag any sections where you
+need more information from the caller with `[NEEDS INPUT: ...]`.

.claude/settings.json

@@ -0,0 +1,50 @@
+{
+  "permissions": {
+    "allow": [
+      "Bash(ls:*)",
+      "Bash(cat:*)",
+      "Bash(echo:*)",
+      "Bash(pwd)",
+      "Bash(date)",
+      "Bash(wc:*)",
+      "Read",
+      "Write",
+      "Edit",
+      "Glob",
+      "Grep",
+      "WebSearch",
+      "WebFetch"
+    ],
+    "deny": [
+      "Bash(rm -rf *)",
+      "Bash(sudo *)",
+      "Bash(chmod 777 *)",
+      "Bash(curl * | bash)",
+      "Bash(wget * | bash)"
+    ]
+  },
+  "hooks": {
+    "PreToolUse": [
+      {
+        "matcher": "Bash",
+        "hooks": [
+          {
+            "type": "command",
+            "command": "bash hooks/pre-tool-use.sh"
+          }
+        ]
+      }
+    ],
+    "PostToolUse": [
+      {
+        "matcher": "Bash",
+        "hooks": [
+          {
+            "type": "command",
+            "command": "bash hooks/post-tool-use.sh"
+          }
+        ]
+      }
+    ]
+  }
+}

.claude/skills/daily-briefing.md

@@ -0,0 +1,40 @@
+---
+name: daily-briefing
+description: Generate a morning briefing with weather, calendar summary, and top news for your focus areas. Demonstrates cron/automation capability.
+---
+
+# Daily Briefing
+
+Generate a concise morning briefing. This skill demonstrates how
+Claude Code can replicate OpenClaw's HEARTBEAT.md autonomous tasks.
+
+## Steps
+
+1. Search for today's top AI news (3 items max)
+2. Search for any breaking tech news
+3. Summarize in a structured briefing
+
+## Output format
+
+```
+## Morning Briefing - [date]
+
+### Top AI News
+- [headline]: [one-sentence summary] ([source])
+- ...
+
+### Action Items
+- [anything that needs attention today]
+
+### Weather
+[Not available without location API - add your own MCP server]
+```
+
+## Automation
+
+To run this daily, see `automation/README.md` for cron and
+launchd setup instructions. You can schedule this with:
+
+```bash
+claude -p "Run /daily-briefing" --output-format text
+```

.claude/skills/send-slack-message.md

@@ -0,0 +1,48 @@
+---
+name: send-slack-message
+description: Send a message to a Slack channel via MCP. Demonstrates multi-channel messaging capability.
+---
+
+# Send Slack Message
+
+Send a message to a Slack channel. This demonstrates how Claude Code
+handles messaging through MCP servers, compared to OpenClaw's 15+
+native channel integrations.
+
+## Prerequisites
+
+You need a Slack MCP server configured in `.mcp.json`. Example:
+
+```json
+{
+  "mcpServers": {
+    "slack": {
+      "command": "npx",
+      "args": ["@anthropic-ai/mcp-slack@latest"],
+      "env": {
+        "SLACK_BOT_TOKEN": "xoxb-your-token-here"
+      }
+    }
+  }
+}
+```
+
+## Usage
+
+Tell Claude Code: "Send a message to #general saying [your message]"
+
+Claude Code will use the Slack MCP server's `send_message` tool.
+
+## How this compares to OpenClaw
+
+OpenClaw has native Slack integration (Bolt SDK, no config needed
+beyond token). Claude Code uses MCP as the integration layer. The
+result is the same: a message arrives in Slack. The setup differs:
+OpenClaw is plug-and-play, Claude Code requires MCP configuration.
+
+## Other messaging channels
+
+- **Telegram:** Native via Claude Code Channels (v2.1.80+)
+- **Discord:** Via MCP server
+- **Email:** Via MCP server (e.g., Gmail MCP)
+- **Teams:** Via MCP server or webhook

.claude/skills/web-research.md

@@ -0,0 +1,50 @@
+---
+name: web-research
+description: Deep web research with source verification. Searches multiple queries, cross-references results, and produces a sourced summary.
+---
+
+# Web Research
+
+Conduct thorough web research on a topic. This skill demonstrates
+Claude Code's built-in WebSearch and WebFetch tools, which match
+OpenClaw's Brave Search + Firecrawl combination.
+
+## How it works
+
+1. Take the user's research question
+2. Generate 3-5 targeted search queries
+3. Search and read the top results for each
+4. Cross-reference claims across sources
+5. Produce a structured summary with citations
+
+## Output format
+
+```
+## Research: [topic]
+
+### Key Findings
+1. [finding] ([source URL])
+2. [finding] ([source URL])
+3. [finding] ([source URL])
+
+### Details
+[Longer explanation with inline citations]
+
+### Sources
+- [title] - [URL]
+- [title] - [URL]
+
+### Confidence
+[High/Medium/Low] - [reason]
+
+### Gaps
+[What could not be verified or found]
+```
+
+## Comparison to OpenClaw
+
+OpenClaw uses Brave Search API with Firecrawl as anti-bot fallback.
+Claude Code uses WebSearch (built-in, no API key needed) and WebFetch
+for full page reading. Both achieve the same outcome. Claude Code's
+advantage: zero configuration. OpenClaw's advantage: configurable
+search provider and caching.