Ashby Automation

Overview of Ashby Automation skill

What Ashby Automation does

Ashby Automation is a Claude skill for operating an Ashby ATS account through natural language, using the Rube MCP server to expose Ashby tools inside Claude Code. It is built for recruiting operations, talent teams, founders, and hiring managers who want to create candidates, search pipelines, manage jobs, handle applications, view interview schedules, and add notes without switching constantly between terminal, browser, and ATS screens.

Best fit for recruiting workflows

The strongest use case is structured recruiting work where the assistant can map a clear instruction to an Ashby action: “create this candidate,” “find candidates matching this email,” “list open engineering jobs,” or “add a note to this application.” Ashby Automation for Recruiting is most useful when your team already uses Ashby as the source of truth and wants faster execution, lookup, and update workflows from Claude Code.

What makes this skill different

Unlike a generic recruiting prompt, the Ashby Automation skill is oriented around actual Ashby toolkit actions such as candidate creation, candidate search, candidate updates, job listing, application management, interview schedule access, and note handling. The important differentiator is not prose generation; it is tool-backed ATS operation through MCP. That means adoption depends on authentication, permissions, and precise inputs more than on prompt style alone.

Important adoption constraints

This skill requires the Rube MCP server and an authenticated Ashby connection. It is not a standalone scraper, resume parser, or replacement for Ashby permissions. If the connected Ashby account cannot access a candidate, job, application, or interview, the skill should not be expected to retrieve or modify it. Teams with strict recruiting compliance rules should verify auditability and approval practices before using it for bulk changes.

How to Use Ashby Automation skill

Ashby Automation install and setup context

Install the skill from the Composio skills repository in your Claude skills environment, then configure MCP access for Rube. A typical directory install command is:

npx skills add ComposioHQ/awesome-claude-skills --skill "Ashby Automation"

The upstream skill itself specifies the required MCP dependency: add the Rube MCP server to Claude Code with https://rube.app/mcp, then authenticate the Ashby account when Claude provides the connection link. Read composio-skills/ashby-automation/SKILL.md first; this repository path has no extra README.md, scripts, rules, or reference folders, so the skill file is the main operating guide.

Inputs the skill needs to act reliably

For create and update actions, provide the exact fields Ashby expects. For candidates, include full name and, when available, email, phone number, LinkedIn URL, GitHub URL, or website. For searches, prefer email for exact matching and names for broader lookup. For job or application work, include the job title, candidate identity, application context, and the intended action. Ambiguous prompts such as “update Sam’s application” are likely to require follow-up because multiple candidates or jobs may match.

Turning a rough goal into a strong prompt

Weak prompt: “Add Jane to Ashby.”

Stronger Ashby Automation usage prompt: “Use Ashby to create a candidate named Jane Smith. Email: [email protected]. LinkedIn: https://linkedin.com/in/janesmith. If a candidate with this email already exists, show me the existing record instead of creating a duplicate.”

The stronger version gives a unique identifier, a desired duplicate-handling policy, and enough structured data for the tool call. For sensitive workflow changes, add an approval checkpoint: “Before changing the application stage or adding a public-facing note, summarize the intended update and ask for confirmation.”

Suggested workflow for first use

Start with read-only actions: list candidates, search by email, retrieve candidate information, and list jobs. Once authentication and permissions are confirmed, move to low-risk writes such as creating a test candidate or adding a clearly labeled internal note. Only then use the skill for operational changes like updating candidate records or managing applications. This staged rollout helps distinguish prompt issues from MCP authentication, Ashby permission, or data-matching problems.

Ashby Automation skill FAQ

Is Ashby Automation only for recruiters?

No. Recruiters are the clearest fit, but recruiting coordinators, hiring managers, founders, and RevOps-style operators can also benefit if they work in Ashby. The skill is especially useful for people who already know what record they need to create or retrieve but want to avoid manual ATS navigation.

How is this better than an ordinary Claude prompt?

An ordinary prompt can draft outreach, summarize interview feedback, or suggest hiring process improvements, but it cannot safely operate Ashby by itself. The Ashby Automation skill is designed around tool-backed actions through Rube MCP, so Claude can call Ashby-related functions after authentication instead of merely giving instructions you must perform manually.

What should I check before installing?

Confirm that your Claude Code setup can use MCP servers, your organization allows Rube MCP connectivity, and your Ashby account has the permissions needed for the workflows you want. Also inspect SKILL.md in the GitHub repository to see the current tool names and examples, because this skill is documented in a single primary file rather than a large support-file package.

When should I not use this skill?

Do not use it as a bulk data migration tool without additional validation, logging, and approval controls. It is also a poor fit if your team does not use Ashby, if you only need generic recruiting copywriting, or if your security policy blocks third-party MCP connections. For high-volume updates, test on a small set of records first and require confirmation before writes.

How to Improve Ashby Automation skill

Improve Ashby Automation prompts with identifiers

The most reliable prompts include stable identifiers: candidate email, candidate name, job title, application context, and the exact action requested. When possible, tell the assistant how to resolve duplicates: “search by email first,” “do not create a new record if one exists,” or “return the matching candidates before updating anything.” This reduces accidental duplicate candidates and wrong-record updates.

Add confirmation rules for risky recruiting actions

For better operational safety, use a consistent instruction pattern: read first, summarize, then ask before writing. This matters for stage changes, notes, candidate updates, and application management. A good prompt is: “Find the candidate by email, show the current application and job association, then ask me to confirm before adding the note.” This keeps Ashby Automation useful without turning natural language into uncontrolled ATS mutation.

Handle common failure modes

Common blockers include expired Ashby authentication, insufficient Ashby permissions, ambiguous candidate names, missing job context, and over-broad list requests. If a command fails, narrow the query: search by exact email, reduce list size, name the job explicitly, or ask Claude to show the tool call it intends to make before executing. The source skill notes that candidate listing may be paginated, so avoid expecting one prompt to return an entire large talent database.

Iterate after the first output

After Claude returns a result, verify the record identity before continuing. If the answer is incomplete, ask for the next specific step rather than restarting broadly: “Use that candidate ID to retrieve full candidate info,” “now list applications for this candidate,” or “add this internal note to the confirmed application.” Treat the first output as record discovery, then use follow-up prompts for controlled recruiting operations.