bonsai-automation
by ComposioHQbonsai-automation helps Claude automate Bonsai operations through Composio Rube MCP. Install the skill, connect Rube, verify the Bonsai connection, then use schema-first tool discovery before running workflows.
This skill scores 66/100, which means it is acceptable for listing but best suited to users already using Rube MCP and Bonsai. It provides enough trigger and setup guidance for an agent to start correctly, but directory users should understand that most operational detail is delegated to live tool discovery rather than embedded examples or supporting documentation.
- Clear trigger and scope: it is explicitly for automating Bonsai operations through Composio's Bonsai toolkit via Rube MCP.
- Operational prerequisites are stated, including Rube MCP availability, an active Bonsai connection, and the requirement to call RUBE_SEARCH_TOOLS first for current schemas.
- The setup and discovery pattern gives agents a repeatable path: verify Rube, manage the Bonsai connection, search tools, then execute using returned schemas.
- No support files, scripts, reference docs, or install command are provided; the skill relies entirely on the single SKILL.md and external Rube/Bonsai schemas.
- Workflow guidance is mostly generic tool discovery and connection setup, with limited concrete Bonsai task examples or edge-case handling in the repository evidence.
Overview of bonsai-automation skill
What bonsai-automation does
bonsai-automation is a Claude skill for running Bonsai business operations through Composio’s Rube MCP server. It is designed for users who want an agent to discover current Bonsai tool schemas, verify the Bonsai connection, and execute tasks through the bonsai toolkit instead of guessing API fields from memory.
The skill’s most important instruction is procedural: always call RUBE_SEARCH_TOOLS first. That matters because Rube tool names, inputs, and execution plans can change, and Bonsai workflows often depend on current schema details.
Best-fit users and workflow automation use cases
The best fit is someone already using Bonsai and willing to connect it through Rube MCP. Typical bonsai-automation for Workflow Automation use cases include asking an agent to inspect available Bonsai actions, prepare a safe execution plan, and then run client, project, invoice, task, or related business-management operations only after confirming the active connection and required fields.
This is not a generic Bonsai how-to guide. It is most useful when your goal is “make Claude operate Bonsai through approved MCP tools” rather than “explain how Bonsai works.”
Key differentiator: schema-first execution
A normal prompt may ask Claude to “create an invoice in Bonsai,” but the model may invent tool arguments or use stale assumptions. The bonsai-automation skill reduces that risk by forcing a discovery pattern:
- Search available Rube tools for the specific Bonsai task.
- Check that the
bonsaitoolkit connection is active. - Read the returned schema and pitfalls.
- Execute only with fields that match the current tool definition.
That schema-first pattern is the main reason to install the skill.
How to Use bonsai-automation skill
bonsai-automation install and MCP setup
Install the skill from the Composio skill collection with:
npx skills add ComposioHQ/awesome-claude-skills --skill bonsai-automation
The skill requires Rube MCP, not a local script bundle. Add https://rube.app/mcp as an MCP server in your client configuration, then confirm the MCP tools are visible. The repository’s only required source file is composio-skills/bonsai-automation/SKILL.md, so read that first; there are no extra scripts/, resources/, or rules/ folders to inspect.
Before expecting useful output, verify:
RUBE_SEARCH_TOOLSis available.RUBE_MANAGE_CONNECTIONScan manage toolkit connections.- The
bonsaiconnection isACTIVE. - If inactive, complete the returned auth flow before running Bonsai actions.
Practical bonsai-automation usage pattern
A reliable bonsai-automation usage flow is:
- Tell the agent the exact Bonsai outcome you want.
- Ask it to call
RUBE_SEARCH_TOOLSfor that specific task. - Ask it to summarize the available tool slug, required fields, optional fields, and known pitfalls.
- Have it check the Bonsai connection with
RUBE_MANAGE_CONNECTIONS. - Provide missing business details.
- Approve execution only after the plan matches your intent.
Example strong prompt:
“Use bonsai-automation to help create a Bonsai invoice for client Acme Ltd for the completed website audit. First search Rube tools for the current Bonsai invoice schema, then check that the bonsai connection is active. Before executing, show me the required fields, any assumptions, and the exact values you plan to send.”
This works better than “make me an invoice” because it supplies intent, entity names, approval requirements, and the required discovery behavior.
Inputs that improve output quality
For best results, provide operational details the tool cannot infer safely:
- Bonsai object type: client, project, invoice, proposal, task, payment, or time entry.
- Names or IDs already known.
- Dates, currency, line items, quantities, rates, tax treatment, and due dates.
- Whether the agent should only draft, search, update, or execute.
- Confirmation rules, such as “ask before sending,” “do not modify existing invoices,” or “show me matching records first.”
If you do not know the exact Bonsai IDs, ask the agent to search first and present candidates. Avoid asking it to update or delete records based only on a name unless the search result is unambiguous.
Repository files to read first
Start with SKILL.md at composio-skills/bonsai-automation/SKILL.md. The important sections are Prerequisites, Setup, Tool Discovery, and Core Workflow Pattern. The file is concise, but the adoption-critical detail is easy to miss: tool discovery is not optional. Any agent workflow that skips RUBE_SEARCH_TOOLS is no longer following the skill’s safety model.
bonsai-automation skill FAQ
Is bonsai-automation useful for beginners?
Yes, if you are comfortable connecting Rube MCP and authorizing Bonsai. The skill gives beginners a safer path than writing raw API calls because it relies on Rube’s current schemas. However, it still assumes you understand the business action you are asking for. The agent can discover fields, but it cannot decide your invoice terms, client identity, or approval policy without input.
How is it different from an ordinary Bonsai prompt?
An ordinary prompt depends on the model’s general knowledge. bonsai-automation instructs the agent to use live MCP discovery through RUBE_SEARCH_TOOLS, then manage the Bonsai connection through RUBE_MANAGE_CONNECTIONS. That makes it better suited for real workflow automation where wrong fields, stale schemas, or inactive authentication would block execution.
When should I not use bonsai-automation?
Do not use it if you only need Bonsai product advice, accounting guidance, or a manual UI walkthrough. Also avoid it for high-risk bulk changes unless you add strict review steps, record matching rules, and confirmation gates. The skill is an execution helper, not a substitute for business review.
Does it require Composio or Rube MCP?
Yes. The skill is built around Composio’s Bonsai toolkit exposed through Rube MCP. If your client cannot connect to https://rube.app/mcp or cannot expose RUBE_SEARCH_TOOLS and RUBE_MANAGE_CONNECTIONS, the core workflow will not run as intended.
How to Improve bonsai-automation skill
Improve bonsai-automation prompts with execution constraints
The fastest way to improve bonsai-automation results is to separate discovery, planning, and execution. For example:
“Search Bonsai tools for updating a project status. Do not execute yet. Show required fields, matching project candidates, and the proposed update payload. Ask for confirmation before running the tool.”
This prevents the agent from moving too quickly from vague intent to live changes.
Common failure modes to watch for
The most common adoption blockers are inactive Bonsai authorization, skipped tool discovery, ambiguous record matching, and missing required fields. If the agent stalls, ask it to report exactly which step failed: MCP availability, Bonsai connection status, schema discovery, missing input, or execution error. That diagnosis is more useful than retrying the same broad request.
Iterate after the first tool result
After the first output, refine based on evidence from Rube rather than assumptions. If a schema includes unexpected required fields, provide those values explicitly. If search returns multiple Bonsai records, choose one by ID or distinguishing detail. If the execution plan includes a risky action, ask for a read-only lookup first.
What would make the skill stronger
The upstream skill would be stronger with example task prompts for common Bonsai operations, explicit read-before-write guidance, and sample confirmation gates for invoices, payments, and record updates. Until those are added, users should build their own checklist around RUBE_SEARCH_TOOLS, RUBE_MANAGE_CONNECTIONS, schema review, and human approval before execution.