detrack-automation

Overview of detrack-automation skill

What detrack-automation does

detrack-automation is a workflow automation skill for operating Detrack through Composio’s Rube MCP tool layer. Its main purpose is not to hard-code one Detrack action, but to help an AI agent discover the current Detrack tool schemas, verify the user’s Detrack connection, and then execute delivery, job, proof-of-delivery, or related Detrack operations through the available Rube tools.

Best fit for Detrack workflow automation

Use the detrack-automation skill if you already use Detrack for delivery operations and want an AI assistant to help run repeatable tasks through MCP instead of manually navigating APIs or dashboards. It is best for users who can provide concrete operational intent, such as “find today’s failed deliveries,” “create delivery jobs from these rows,” or “check POD status for these tracking numbers.”

Key differentiator: schema-first execution

The important design choice in this skill is that it instructs the agent to call RUBE_SEARCH_TOOLS before running Detrack actions. That matters because Composio tool schemas can change, and guessing parameters is risky for logistics workflows. A good detrack-automation guide should therefore prioritize discovery, connection status, and tool-specific input validation before execution.

Adoption considerations

This is a thin, MCP-dependent skill with one main source file: SKILL.md. There are no bundled scripts, rule files, or local templates to adapt. That makes detrack-automation lightweight, but also means your results depend heavily on having Rube MCP configured, an active Detrack connection, and prompts that include the real business constraints behind the task.

How to Use detrack-automation skill

detrack-automation install and setup context

Install the skill from the Composio skills repository in a compatible skills-enabled client:

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

Then configure Rube MCP by adding https://rube.app/mcp as an MCP server in your client. The skill expects RUBE_SEARCH_TOOLS to be available. Before using Detrack actions, call RUBE_MANAGE_CONNECTIONS for the detrack toolkit and complete the returned authentication flow if the connection is not ACTIVE.

Inputs the skill needs from you

For reliable detrack-automation usage, give the agent more than a vague instruction. Include:

• The Detrack object or workflow: jobs, deliveries, PODs, tracking, drivers, recipients, or statuses.
• The action: search, create, update, cancel, export, reconcile, or summarize.
• Identifiers: tracking numbers, job IDs, order IDs, dates, customer names, or depot/route details.
• Safety limits: read-only first, ask before updates, process only a sample, or confirm before bulk changes.
• Output format: table, CSV-ready rows, exception list, operational summary, or next-step checklist.

Weak prompt: “Use Detrack to update deliveries.”

Stronger prompt: “Using detrack-automation, first discover the current Detrack tools. Check my active Detrack connection. Then find deliveries scheduled for 2025-02-14 with status failed or partial, summarize the failure reasons by route, and do not update anything without confirmation.”

Practical workflow for agents

A good detrack-automation for Workflow Automation run usually follows this sequence:

1. Read SKILL.md to confirm the required MCP tools and workflow pattern.
2. Call RUBE_SEARCH_TOOLS with the specific Detrack use case, not a generic query.
3. Check the Detrack connection with RUBE_MANAGE_CONNECTIONS.
4. Select the exact tool slug and schema returned by Rube.
5. Map the user’s request into the discovered schema.
6. Run read operations first when possible.
7. Ask for confirmation before destructive, bulk, or customer-visible updates.
8. Return both the result and the tool assumptions used.

Repository files to read first

Start with composio-skills/detrack-automation/SKILL.md. It contains the complete operational contract: prerequisites, setup, tool discovery, connection checking, and the core execution pattern. Because this skill has no extra README.md, rules/, resources/, or scripts, there is little hidden behavior to inspect. The value is in following the MCP discovery pattern precisely.

detrack-automation skill FAQ

Is detrack-automation useful without Rube MCP?

No. The detrack-automation skill requires Rube MCP because it relies on RUBE_SEARCH_TOOLS and RUBE_MANAGE_CONNECTIONS. If your client cannot connect to MCP servers or expose Rube tools to the agent, this skill will not execute Detrack operations.

How is this better than an ordinary Detrack prompt?

A normal prompt may describe what you want, but it may guess API fields or invent tool names. The detrack-automation skill tells the agent to discover current Composio Detrack tools first, inspect schemas, confirm connection state, and only then execute. That reduces brittle automation and is especially important when creating or updating logistics records.

Is the detrack-automation skill beginner-friendly?

It is beginner-friendly if your MCP client is already set up and you understand the Detrack task you want performed. It is less beginner-friendly if you expect a packaged app, visual workflow builder, or prebuilt delivery templates. The skill gives an agent a safe execution pattern; it does not replace Detrack domain knowledge.

When should I not use this skill?

Do not use detrack-automation for unsupported Detrack actions that are not returned by RUBE_SEARCH_TOOLS, for offline batch processing without an active connection, or for high-risk bulk updates without a review step. If you need custom validation, audit logs, approval workflows, or transformations from complex spreadsheets, pair the skill with your own process rules.

How to Improve detrack-automation skill

Make detrack-automation prompts more specific

The most common failure mode is under-specified intent. Improve prompts by naming the business object, date range, identifiers, status filters, and expected output. Instead of “check deliveries,” say “search deliveries created this week for customer ACME where status is not delivered, return tracking number, recipient, last event, failure reason, and recommended follow-up.”

Add guardrails for risky operations

For create, update, cancel, or bulk workflows, explicitly require a preview before execution. Good guardrails include “read-only until I approve,” “limit to 10 records for the first run,” “show the exact fields that will change,” and “skip records with missing recipient phone or address.” These constraints help the agent use discovered Detrack schemas without making unsafe assumptions.

Iterate after the first output

After the first run, ask the agent to refine based on operational quality, not just formatting. Useful follow-ups include: “separate driver-caused issues from customer-unavailable issues,” “group exceptions by route,” “turn this into a CSV import shape,” or “rerun only for jobs missing POD.” Iteration works best when you preserve the Rube search session or ask the agent to rediscover tools if the schema is uncertain.

Improve the skill source for team use

If your team depends on detrack-automation, consider extending the local skill copy with examples for your common Detrack workflows, standard status definitions, approval rules, and output templates. The upstream skill is intentionally compact; adding organization-specific examples can make it faster and safer without changing the core requirement to search tools before execution.