Ashby MCP Server
Click on "Deploy Server".
Wait a few minutes for the server to deploy. Once ready, it will show a "Started" state.
In the chat, type
@followed by the MCP server name and your instructions, e.g., "@Ashby MCP Server List my open jobs and how many active applications each has"
That's it! The server will respond to your query, and you can continue using it as needed.
Here is a step-by-step guide with screenshots.
Ashby MCP server through Cotlet.dev
Connect Ashby to ChatGPT, Claude, Codex and Claude Code through Cotlet.dev. Work with jobs, candidate profiles, applications, interviews and offers in your signed-in Ashby account.
This is an unofficial integration, not affiliated with Ashby. Cotlet.dev hosts the connection through your Cloud Browser; this repository provides an optional Python stdio MCP bridge.
Connect your account
Open Ashby in your Cotlet.dev Cloud Browser, sign in yourself and enable it for agents.
Add Cotlet.dev to your agent using the connection details below.
ChatGPT
Open Settings > Apps & Connectors > Create. Use https://cotlet.dev/mcp/v1/connect as the MCP server URL and choose OAuth.
Claude
Open Settings > Connectors > Add custom connector and enter https://cotlet.dev/mcp/v1/connect.
Codex
Create a Cotlet.dev API key in your console and set COTLET_API_KEY in your environment. Add this to your Codex configuration:
[mcp_servers.cotlet]
url = "https://cotlet.dev/mcp/v1"
bearer_token_env_var = "COTLET_API_KEY"Claude Code
claude mcp add --transport http cotlet https://cotlet.dev/mcp/v1 --header "Authorization: Bearer $COTLET_API_KEY"Related MCP server: Pandapé MCP
Recruiting tasks
Examples based on the live Ashby integration reference:
List open jobs and compare active applications.
Find candidates by name or email, then read their applications, notes and interviews.
List applications in a job stage and identify those waiting longest.
Schedule a technical screen with selected interviewers and invite settings.
Draft or send a candidate email using an available template.
Archive an application with a configured reason and an optional rejection email.
Create an offer from your organization's fields and send its letter after required approvals.
The live page documents these 12 operations:
list_jobssearch_candidatesget_candidatelist_applicationslist_interviewslist_archive_reasonslist_email_templatesschedule_interviewsend_emailarchive_applicationcreate_offersend_offer_letter
Hosted MCP exposes these operations through ashby_read and ashby_write. Writes ask for confirmation by default; the live reference describes per-tool settings, roles and offer approval requirements. Scheduling and offers need appropriate Ashby permissions.
Optional local stdio bridge
Python 3.10 or newer is required, with no runtime dependencies. Clone this repository and configure a client that supports stdio:
{"mcpServers":{"ashby":{"command":"python3","args":["/absolute/path/to/ashby-mcp/ashby_agent.py"],"env":{"COTLET_API_KEY":"your-cotlet-key"}}}}The bridge lists seven browser helpers: ashby_guide, ashby_status, ashby_open, ashby_read, ashby_click, ashby_type and ashby_close. These forward to Cotlet.dev's shared browser tools. They are separate from the 12 hosted Ashby operations above. Call the guide first and close the browser when finished. Tool discovery works without a key; executing a tool requires your Cotlet.dev key and connected account.
python3 ashby_agent.py connect
python3 ashby_agent.py guide
python3 ashby_agent.py status
python3 ashby_agent.py serveYou can also build the included Dockerfile and run it with COTLET_API_KEY supplied as an environment variable.
Reference and access
Licensed under Elastic License 2.0. See LICENSE.
Available Tools
7 toolsashby_clickB
Click a numbered element on the current Ashby page.
Inputs: element (required).
ashby_closeA
Close the Ashby browser session when the work is done, so credits stop running.
Inputs: no named parameters.
ashby_guideA
How to work in Ashby (jobs, candidates, interviews and offers): its screens and the sign-in link. Call this first.
Inputs: task.
ashby_openA
Open a Ashby page in your signed-in Cloud Browser (default https://app.ashbyhq.com) to work with jobs, candidates, interviews and offers.
Inputs: url.
ashby_readB
Read the current Ashby page: its text and numbered buttons, links and fields.
Inputs: element.
ashby_statusA
Check that your Cotlet.dev account and Ashby are ready: account, credits and Ashby connection.
Inputs: no named parameters.
ashby_typeC
Type text into a numbered field on the current Ashby page.
Inputs: text (required), element (required).
Tool Schema Changelog
Recent tool additions, removals, and schema changes observed during successful MCP inspections.
Loading tool schema changelog…
TDQS
Scored across 7 tools
Each tool maps to a clearly distinct step in a browser session lifecycle: guide (instructions), status (readiness), open (navigate), read (inspect), click (interact), type (input), close (teardown). The two informational tools (guide vs status) are the closest pair, but their descriptions clearly separate onboarding guidance from account/connection verification.
All tools use the same ashby_ prefix followed by a single verb in snake_case (open, read, click, type, close, guide, status). No mixed conventions or vague multi-word names; the pattern is immediately predictable.
Seven tools is well-scoped for a single-site browser automation server: one tool per primitive action plus setup/teardown and help. Nothing feels redundant or padded, and nothing essential to the loop is missing at this count.
The core loop (open → read → click/type → close) plus session management and help covers the stated purpose. Minor gaps exist, such as scrolling/navigation back, waiting for page loads, or keyboard submit (Enter), which agents may need on long or dynamically loading Ashby pages, but these are workable through read+click.
Maintenance
Related MCP Connectors
Connect Claude, Cursor, Codex, and other MCP clients to manage your Cavuno job board.
Review candidates and run hiring actions in InterviewFlowAI from Claude, ChatGPT, Codex and more.
Create contacts, move deals, send proposals and invoices, read your inbox, from Claude or ChatGPT.
Connect authorized ad accounts to AI assistants for audits, reporting, and approval-gated changes.
Related MCP Servers
- AlicenseAqualityFmaintenanceEnables AI tools like Claude and Codex to access and manage Recruit CRM data including candidates, jobs, companies, tasks, meetings, notes, and call logs through natural language.699 npmMIT
- FlicenseAqualityDmaintenanceConnects Claude to the Pandapé recruitment API, enabling natural language management of vacancies, structured candidate review against custom criteria, pipeline stage updates, and CV downloads.14-
- FlicenseNot gradedqualityDmaintenanceEnables Claude to securely search, read, and update your ATS data — candidates, mandates, pipelines, tasks, and interviews — through natural language, with configurable write permissions and confirmation for actions like scheduling and emailing.-

hirly MCP serverofficial
AlicenseNot gradedqualityCmaintenanceConnects an AI assistant to a job-matching account so it can read roles matched to your resume, pull full job details and hiring contacts, and save tailored resume versions. With explicit opt-in and per-job confirmation in the chat, it can also start applications, relay questions only you can answer, and report each application's status.MIT