metaphor-automation

Overview of metaphor-automation skill

What metaphor-automation does

metaphor-automation is a Claude skill for running Metaphor tasks through Composio’s Rube MCP layer. Instead of hard-coding old tool names or schemas, it instructs the agent to discover the current Metaphor tools first, verify the user’s Metaphor connection, then execute the selected operation with the schema returned by Rube.

This makes the metaphor-automation skill most useful when you want an AI agent to operate against Metaphor reliably without guessing which MCP tool exists today.

Best fit for Workflow Automation users

Use metaphor-automation for Workflow Automation when your agent needs to call Metaphor as part of a repeatable process: searching, retrieving, enriching, or acting on Metaphor-managed data through Composio. It is best for users already working with MCP-enabled clients and who can authorize a Metaphor connection through Rube.

It is less useful as a standalone prompt library. The value comes from live MCP access: RUBE_SEARCH_TOOLS, RUBE_MANAGE_CONNECTIONS, and the Metaphor toolkit exposed through Composio.

Key differentiator: schema discovery first

The strongest design choice in this skill is its insistence that the agent call RUBE_SEARCH_TOOLS before taking action. That matters because MCP tool schemas can change, available actions may vary by account, and stale examples can produce failed calls.

For install-decision purposes, this is a practical guardrail: metaphor-automation is not trying to memorize Metaphor’s API. It teaches the agent to ask Rube for the current tool list, input schema, execution plan, and pitfalls before running the workflow.

What to check before installing

Before installing, confirm your environment can use Rube MCP and that you are comfortable authorizing a Metaphor toolkit connection. The upstream skill is compact and centered on SKILL.md; there are no bundled scripts, examples, resources, or test fixtures. That keeps adoption simple, but it also means your prompts must provide the task details the skill cannot infer.

How to Use metaphor-automation skill

metaphor-automation install context

Install the skill from the repository path used by your skill manager, for example:

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

Then add Rube MCP to your AI client using:

https://rube.app/mcp

The skill requires the rube MCP server. In practice, verify that RUBE_SEARCH_TOOLS is available before expecting metaphor-automation usage to work. If the MCP server is missing, the skill can describe a workflow but cannot execute Metaphor actions.

Required setup before the first run

A successful run usually follows this order:

1. Confirm RUBE_SEARCH_TOOLS responds.
2. Use RUBE_MANAGE_CONNECTIONS with toolkit metaphor.
3. If the connection is not ACTIVE, open the returned authorization link.
4. Re-check the connection status before asking the agent to execute the task.
5. Ask the agent to search tools for your specific Metaphor use case, not just “Metaphor operations.”

The important adoption blocker is connection state. Many failures will not be prompt-quality problems; they will be missing MCP access or an inactive Metaphor connection.

Turn a rough goal into a usable prompt

Weak prompt:

Use Metaphor to find some data.

Stronger prompt:

Use the metaphor-automation skill. First call RUBE_SEARCH_TOOLS for the exact use case: “find recent Metaphor records matching [topic/entity/filter] and return the top results with source fields.” Check the Metaphor connection with RUBE_MANAGE_CONNECTIONS. Use only the schema returned by Rube. Before executing, summarize the chosen tool, required inputs, optional filters, and any missing information you need from me.

This improves output because it gives the agent the operation type, entity or topic, expected result shape, and a rule not to invent fields. For automation, add limits, date ranges, deduplication rules, output format, and what should happen if no results are found.

Repository files to read first

Start with composio-skills/metaphor-automation/SKILL.md. It contains the whole skill: prerequisites, setup, tool discovery, connection checks, and the core workflow pattern. There are no separate README.md, scripts/, references/, or resources/ folders in the current skill path, so do not expect hidden implementation details elsewhere.

For deeper product context, review Composio’s Metaphor toolkit documentation linked from the skill: https://composio.dev/toolkits/metaphor.

metaphor-automation skill FAQ

Is metaphor-automation enough without Rube MCP?

No. The metaphor-automation skill depends on Rube MCP being available in the client. Without RUBE_SEARCH_TOOLS, the most important behavior—discovering current Metaphor tools and schemas—cannot happen. Without RUBE_MANAGE_CONNECTIONS, the agent also cannot verify whether the Metaphor toolkit is authorized.

How is this better than an ordinary prompt?

An ordinary prompt may ask the model to “use Metaphor,” but it may hallucinate tool names, outdated parameters, or unsupported actions. metaphor-automation gives the agent a safer execution pattern: discover tools, check connection, inspect schemas, then run the selected tool. That makes it more reliable for live Workflow Automation than a static instruction alone.

Is this suitable for beginners?

It is beginner-friendly if you already know how to add an MCP server to your AI client. It is not ideal for users who have never configured MCP or OAuth-style tool connections. The skill’s documentation is short, so beginners should be ready to follow the Rube connection flow and ask the agent to explain each returned tool schema before execution.

When should you not use this skill?

Do not use metaphor-automation if you only need general brainstorming about Metaphor, if your client cannot connect to MCP servers, or if your workflow requires local scripts and validations that this skill does not provide. Also avoid it when you need a fixed, audited API wrapper; this skill intentionally relies on live Rube tool discovery.

How to Improve metaphor-automation skill

Improve metaphor-automation inputs

The best results come from precise task framing. Include:

• the exact Metaphor operation you want
• entity names, query terms, filters, or IDs
• expected output fields
• result limits and sort preferences
• whether the agent may take action or only prepare a plan
• fallback behavior when the schema lacks a desired field

This reduces unnecessary tool searches and prevents the agent from treating a broad research request as an executable automation step.

Avoid common failure modes

The most common failures are skipped discovery, inactive connection, guessed schemas, and vague user goals. Tell the agent explicitly:

Do not call any Metaphor tool until RUBE_SEARCH_TOOLS has returned the current schema and RUBE_MANAGE_CONNECTIONS confirms an active Metaphor connection.

If a call fails, ask the agent to compare the attempted input against the schema returned by Rube, not against memory or old examples.

Iterate after the first output

After the first run, improve the workflow by asking for an execution note: selected tool slug, required fields used, optional fields omitted, returned errors, and recommended next call. This turns a one-off run into a reusable metaphor-automation guide for your team.

For repeated workflows, save a successful prompt pattern with your required filters, output format, and approval checkpoints.

Extend the skill safely for your environment

If you maintain a fork, add examples that match your real use cases: search templates, connection-check snippets, expected outputs, and error-handling rules. Keep the core “search tools first” behavior intact. The biggest improvement is not more prose; it is adding organization-specific prompts that make tool discovery more targeted and execution less ambiguous.