wati-automation

Overview of wati-automation skill

What wati-automation does

wati-automation is a Claude skill for automating Wati tasks through Composio’s Rube MCP integration. Instead of hard-coding Wati API calls, the skill tells the agent to discover the currently available Wati tools, confirm the account connection, inspect live input schemas, and then execute the workflow through Rube.

This matters because Wati automation often fails when a prompt assumes outdated field names, missing connection state, or the wrong tool slug. The main job of the wati-automation skill is to reduce that guesswork before the agent tries to send, update, fetch, or manage Wati data.

Best-fit users and workflows

Use this skill if you want an AI agent to help with Wati operations inside a workflow automation setup, especially when the exact Composio tool schema may change over time. It is a good fit for teams using Wati for WhatsApp customer messaging, contact handling, campaign support, support workflows, or operational automations where Claude needs to call real tools rather than only draft instructions.

The best-fit reader is someone who already has, or can authorize, a Wati connection through Rube MCP. If you only need marketing copy, message templates, or a manual checklist, an ordinary prompt may be enough.

Important adoption requirement

The key dependency is Rube MCP. The skill requires RUBE_SEARCH_TOOLS and RUBE_MANAGE_CONNECTIONS, plus an active Wati connection under the wati toolkit. The upstream skill is intentionally small: it does not ship scripts, helper resources, or local code. Its value is in enforcing the correct discovery-first workflow before tool execution.

How to Use wati-automation skill

wati-automation install context

Install the skill from the Composio skills repository in a compatible Claude skills environment:

npx skills add ComposioHQ/awesome-claude-skills --skill wati-automation

Then configure Rube MCP in your client by adding the MCP endpoint:

https://rube.app/mcp

The skill itself does not provide a standalone Wati client. Before expecting useful output, verify that your agent can access RUBE_SEARCH_TOOLS. Then use RUBE_MANAGE_CONNECTIONS for toolkit wati and complete the returned authorization flow if the connection is not active.

Inputs the skill needs from you

For reliable wati-automation usage, give the agent a concrete Wati task, the business goal, and any identifiers or constraints you already know. Strong inputs include:

• The action you want: send a message, find contacts, update data, check a campaign-related object, or inspect available Wati operations.
• Known fields: phone number, contact name, template name, campaign identifier, tag, status, or date range.
• Safety limits: whether the action may send messages, only preview data, update records, or require confirmation first.
• Output format: summary, table, execution plan, dry-run checklist, or final confirmation.

A weak prompt is: “Automate Wati.”
A stronger prompt is: “Using wati-automation, find the current Wati tools, confirm the wati connection is active, then prepare a dry-run plan to send an approved WhatsApp template to these three contacts. Do not execute the send step until I confirm.”

Practical workflow for first run

Start by reading composio-skills/wati-automation/SKILL.md; it is the only source file and contains the required workflow. The agent should first call RUBE_SEARCH_TOOLS with a specific use case, not a generic one. For example, “send approved Wati template message to contact” will usually produce more useful schemas than “Wati operations.”

After tool discovery, the agent should call RUBE_MANAGE_CONNECTIONS with toolkit wati. If the connection is inactive, stop and complete authorization. Once active, use the schema returned by tool discovery to build the exact call. This sequence is the main difference between a dependable wati-automation guide and a brittle prompt that guesses tool names.

Tips for higher-quality execution

Ask the agent to keep the same Rube session when moving from discovery to connection check to execution. This helps preserve tool recommendations and avoids repeating schema discovery unnecessarily. If the workflow can affect customers, require a confirmation step before sending messages or updating Wati records.

For ambiguous requests, tell the agent to ask clarifying questions instead of choosing defaults. For example, if you mention a campaign but provide no audience, template, or send condition, the agent should not invent them.

wati-automation skill FAQ

Is wati-automation for Workflow Automation or just Wati prompts?

wati-automation is for Workflow Automation through Rube MCP. It is not merely a prompt template for writing WhatsApp messages. Its purpose is to help an agent discover and call real Wati tools exposed by Composio, while checking connection state and current schemas before execution.

What does the skill not include?

The repository entry does not include local scripts, sample datasets, custom validation rules, or a separate README. It also does not document every Wati operation directly. Instead, it depends on live tool discovery through RUBE_SEARCH_TOOLS, which is useful when schemas change but means you must have MCP access working.

Is this suitable for beginners?

It is suitable for beginners who are comfortable connecting an MCP server and following an authorization link. It is less suitable if you expect a no-code Wati dashboard replacement. The skill assumes the agent can use Rube tools, manage sessions, and interpret returned schemas.

When should I not use this skill?

Do not use it when you do not have permission to connect the Wati account, when customer-impacting actions cannot be safely delegated, or when your environment cannot access Rube MCP. Also avoid it for purely editorial tasks such as drafting a WhatsApp campaign message; a normal writing prompt is simpler for that.

How to Improve wati-automation skill

Improve wati-automation prompts with specific intent

The biggest improvement comes from replacing broad requests with execution-ready intent. Include the Wati object, target audience, action, timing, and approval rule. For example: “Search available Wati tools for contact lookup by phone number, confirm the connection, retrieve the contact if present, and return the fields required to update its tags. Do not update yet.”

This gives the skill enough context to search the right tool schemas and avoid unsafe assumptions.

Common failure modes to prevent

The most common failure is skipping tool discovery and guessing a tool name or field. Make discovery mandatory in your prompt. Another failure is trying to execute before the Wati connection is active; ask the agent to report connection status before continuing. A third is vague customer-impacting automation, such as “message all leads,” without template, consent, segmentation, or approval constraints.

Iterate after the first output

After the first tool discovery result, refine the workflow using the returned schema. If required fields are missing, provide them rather than asking the agent to improvise. If the result includes multiple possible Wati tools, ask the agent to compare them briefly and choose the safest one for your goal.

For production-like usage, start with read-only actions, then dry runs, then confirmed writes or sends. This staged approach makes wati-automation safer and more predictable without changing the skill itself.

Repository improvement ideas

The upstream skill would become easier to adopt if it added examples for common Wati tasks, such as contact lookup, template-message preparation, campaign inspection, and safe send confirmation. A small troubleshooting section for inactive connections, missing RUBE_SEARCH_TOOLS, and ambiguous schemas would also reduce onboarding friction for new users.