Available tools for Reeve agents, tool profiles, and how to create custom tools.

Agent Tools

Tools are capabilities that agents use during conversations — running commands, searching the web, controlling a browser, sending messages, and more. Reeve provides a rich set of built-in tools, plus connector tools from your linked platforms.

Built-in Tools

File System

Tool	What it does
`read`	Read file contents
`write`	Create or overwrite files
`edit`	Find-and-replace edits in files
`apply_patch`	Multi-hunk structured patches

Execution

Tool	What it does
`exec`	Run shell commands with background support
`process`	Manage background processes (poll, log, kill)

Web

Tool	What it does
`web_search`	Search the web via Brave Search
`web_fetch`	Fetch and extract content from URLs

Browser

Tool	What it does
`browser`	Full browser automation — navigate, snapshot, click, type, screenshot

Communication

Tool	What it does
`message`	Send messages across WhatsApp, Telegram, Discord, Slack, and more

Sessions & Agents

Tool	What it does
`sessions_list`	List active sessions
`sessions_history`	View session transcripts
`sessions_send`	Send a message to another session
`sessions_spawn`	Create a sub-agent session
`session_status`	Check current session state
`agents_list`	List agents available for spawning

Memory

Tool	What it does
`memory_search`	Semantic search over memory files
`memory_get`	Read a specific memory file

Automation

Tool	What it does
`cron`	Manage scheduled jobs and wakeups
`gateway`	Restart, configure, or update the gateway

Devices

Tool	What it does
`nodes`	Control paired devices — camera, screen, notifications
`canvas`	Render UI on paired device screens
`image`	Analyze images with a vision model

Connector Tools

When you connect external platforms via Connectors, agents get access to platform-specific tools:

Tool	Source	What it does
`reeve_shopify`	Shopify	Products, orders, customers, revenue
`reeve_analytics`	PostHog / GA4	Traffic, funnels, top pages
`reeve_ads`	Meta / Google	Campaign performance, ad generation
`reeve_email`	Klaviyo	Email campaigns, flows, segments
`reeve_revenue`	Stripe	MRR, LTV, subscriptions
`reeve_social`	Connected accounts	Social metrics, trends
`reeve_support`	Gorgias / Zendesk	Support tickets
`reeve_brand`	Brand profiler	Brand analysis, competitive intel

Tool Profiles

Profiles restrict which tools an agent can access:

Profile	Tools included
`full`	All tools (default)
`coding`	File system, runtime, sessions, memory, image
`messaging`	Message, sessions list/history/send/status
`minimal`	Session status only

Set a profile globally or per-agent:

{
  "tools": {
    "profile": "coding"
  }
}

Per-agent override:

{
  "agents": {
    "list": [{
      "id": "support-bot",
      "tools": { "profile": "messaging" }
    }]
  }
}

Allow / Deny Lists

Fine-tune beyond profiles with allow and deny lists:

{
  "tools": {
    "profile": "coding",
    "allow": ["web_search", "web_fetch"],
    "deny": ["exec"]
  }
}

Deny always wins over allow. Use group:* shorthands for convenience:

Group	Tools
`group:runtime`	exec, bash, process
`group:fs`	read, write, edit, apply_patch
`group:sessions`	sessions_list, sessions_history, sessions_send, sessions_spawn, session_status
`group:memory`	memory_search, memory_get
`group:web`	web_search, web_fetch
`group:ui`	browser, canvas
`group:automation`	cron, gateway
`group:messaging`	message

Custom Tools via Plugins

Extend the tool set with plugins:

Create a plugin manifest defining your tool's schema
Implement the tool handler
Register the plugin in your config

Plugins can add both tools and CLI commands. See Agent Tools Plugin for the full guide.

Skills

Skills are prompt-level guidance that teach agents how to use tools effectively. While tools define what an agent can do, skills define how it should approach a task.

Built-in skills include:

frontend-design — UI development best practices
test-driven-development — Test-first coding workflows
systematic-debugging — Structured debugging methodology
using-git-worktrees — Git branching strategies

Configure skills per-agent or use them on-demand via slash commands.

Tools are presented to the agent in two ways: as structured function schemas (for the model API) and as human-readable text in the system prompt. If a tool doesn't appear in either, the agent cannot use it.

Agent Tools

On this page