Claude Code

Connect Claude Code to DashClaw and get your first governed action into /decisions in under 20 minutes.

Instance URL detected: https://your-dashclaw-instance.example.com

Watch the 3-minute walkthrough

See the full install → first approval round-trip end to end. Skip if you prefer to follow the written steps below.

Screencast: <SCREENCAST_URL>

Deploy DashClaw

Get a running instance. Click the Vercel deploy button or run locally.

Already have an instance? Skip to Step 3.

Install the hook scripts

One command copies all three governance hooks (PreToolUse, PostToolUse, Stop), the vendored intel module that powers semantic tool classification, and merges the matching settings.json blocks. Re-run after each git pull to upgrade.

Terminal

# From the DashClaw repo root:
npm run hooks:install

# Or from any other project, pointing at a checked-out DashClaw clone:
node /path/to/DashClaw/scripts/install-hooks.mjs --target=.

# Manual fallback (skips settings merge):
mkdir -p .claude/hooks
cp hooks/dashclaw_pretool.py  .claude/hooks/
cp hooks/dashclaw_posttool.py .claude/hooks/
cp hooks/dashclaw_stop.py     .claude/hooks/
cp -r hooks/dashclaw_agent_intel .claude/hooks/

Set environment variables

Claude Code reads these from the shell or a .env file in the project root.

.env

DASHCLAW_BASE_URL=https://your-dashclaw-instance.example.com
DASHCLAW_API_KEY=<your-workspace-api-key>
DASHCLAW_HOOK_MODE=enforce

Add hooks to Claude Code settings

Merge this into your project's .claude/settings.json (or ~/.claude/settings.json for global).

.claude/settings.json

{
  "hooks": {
    "PreToolUse": [
      {
        "matcher": "Bash|Edit|Write|MultiEdit",
        "hooks": [
          {
            "type": "command",
            "command": "python .claude/hooks/dashclaw_pretool.py",
            "timeout": 3600000
          }
        ]
      }
    ],
    "PostToolUse": [
      {
        "matcher": "Bash|Edit|Write|MultiEdit",
        "hooks": [
          {
            "type": "command",
            "command": "python .claude/hooks/dashclaw_posttool.py"
          }
        ]
      }
    ],
    "Stop": [
      {
        "hooks": [
          {
            "type": "command",
            "command": "python .claude/hooks/dashclaw_stop.py"
          }
        ]
      }
    ]
  }
}

Connect Discord (2 minutes)

A Discord bot turns your phone into a one-tap approval surface for risky tool calls. The built-in Discord adapter posts a DM with Approve / Deny buttons when a policy requires human judgment. Telegram parity; ENV-only setup.

.env.local (or Vercel env vars)

DISCORD_BOT_TOKEN=<token>
DISCORD_PUBLIC_KEY=<64-char-hex>
DISCORD_APPROVER_USER_ID=<numeric-user-id>
DISCORD_APPROVER_ORG_ID=<your-org-id>
# Kill switch — leave unset or set to true to enable DMs
# DASHCLAW_ALERTS_DISCORD=false

Step-by-step Discord Developer Portal walkthrough is printed below.

Run Claude Code and trigger a tool call

Ask Claude Code to do anything that uses Bash, Edit, Write, or MultiEdit. The hook fires automatically. For policies that require approval, your phone will DM you.

Example prompt

Create a file called hello.txt with the contents "Hello from a governed agent"

Watch the terminal — you should see [DashClaw] messages as the hook evaluates the action.

See the result in DashClaw

Open your DashClaw dashboard to confirm the action was recorded.

Go to /decisions — you should see your tool call in the ledger with action_type 'other' (for a simple file write) or 'security' (for sensitive files), status 'completed'. Approvals that ran through Discord show approved_by starting with 'discord:'.

What success looks like

Go to /decisions — you should see your Claude Code tool call in the ledger. Look for action_type 'other' or 'security' with agent_id 'claude-code' and status 'completed'.

Navigate to /decisions in your DashClaw instance. Your action should appear in the ledger within seconds of the agent run.

Governance as Code

Drop a guardrails.yml in your project root to enforce policies without code changes. DashClaw evaluates these rules at the guard step before any action executes.

guardrails.yml

version: 1
project: my-claude-code-project
description: >
  Governance policy for Claude Code tool calls.
  Blocks destructive shell commands. Warns on deployment.

policies:
  - id: block_destructive_shell
    description: Block rm -rf and database drops
    applies_to:
      tools:
        - Bash
    rule:
      block: true
    when:
      command_contains:
        - "rm -rf"
        - "drop table"

  - id: warn_on_deploy
    description: Require approval for deployment commands
    applies_to:
      tools:
        - Bash
    rule:
      require: approval
    when:
      command_contains:
        - "git push"
        - "vercel deploy"

Discord setup

## Discord Developer Portal walkthrough

### Create the bot
- Open https://discord.com/developers/applications -> New Application
- Name the app; skip the Installation tab
- Open the "Bot" tab -> Reset Token -> copy as DISCORD_BOT_TOKEN
- Open "General Information" -> copy the Public Key as DISCORD_PUBLIC_KEY
- Under "Privileged Gateway Intents" leave ALL off (button-only bot)

### Invite the bot to a mutual server (so DMs work)
- Open "OAuth2" -> URL Generator -> scopes: "bot" -> permissions: "Send Messages"
- Paste the URL in a browser, invite the bot to a personal test server
- In Discord client, enable Developer Mode (Settings -> Advanced)
- Right-click your own user in the member list -> Copy User ID
- Paste as DISCORD_APPROVER_USER_ID

### Register the interactions endpoint
- In "General Information" set:
  Interactions Endpoint URL: https://<your-deployment>/api/discord/interactions
- Discord sends a PING; DashClaw responds {type:1} and the URL saves.

### Verify
- Trigger a Claude Code tool call that hits an approval-required policy
- Your phone's Discord app lights up; tap Approve or Deny
- The DM edits in place to show APPROVED or DENIED with timestamp

Watch the 3-minute walkthrough

End-to-end install to first Discord approval. Published on Loom / YouTube (Unlisted) — backfilled by plan 02-01 once recorded.

Screencast: <SCREENCAST_URL>