Claude Code searches and submits job applications
Artur Erofeev used Claude Code with a real Chrome session to collect jobs, tailor application materials and submit forms, retaining CSV review and manual handling of authentication barriers.
Original prompt
Shared by the source author. See source notes for context.
Find jobs based on my resume
Reach people exploring AI agents
See placements →Evidence at a glance
Sourced account- Source author
- Artur Erofeev
- Original account
- Read the source ↗
- Published
- 17 Mar 2026
- Source link checked
- 15 Oct 2026
- Reported result
- Erofeev reports repeated tailored application submissions with reusable scripts and tracking logs, while authentication barriers still require human intervention.
- Human involvement
- Erofeev curates the job list, spot-checks letters and handles CAPTCHAs and two-factor steps; no interview outcome is established.
A linked account supports attribution. It does not mean we independently reproduced its result.
Report an issue with this case
Open correction emailOpens your email app with the case link included. Nothing is sent automatically.
How the agent did it
- Read the resume and search LinkedIn for relevant roles.
- Collect and deduplicate a CSV that the owner reviews.
- Read each selected description, tailor materials and fill the application.
Artur Erofeev built a job-application workflow around Claude Code, Chrome remote debugging and a Markdown skill. The agent reads a resume, searches LinkedIn and collects matches into a CSV, deduplicating companies already listed. Erofeev reviews that list before application work begins.
For each selected role, the agent reads the description, prepares a cover letter from the applicant's actual experience, fills the application and records whether it was submitted or skipped. It writes Selenium scripts while inspecting unfamiliar forms and saves successful patterns for reuse. The account describes handling several application systems and recovering from stale elements and intercepted clicks.
The workflow still has clear limits. CAPTCHAs and two-factor checks require human handling. Unusual screening questions can need judgment, and Erofeev spot-checks letters. He also reports changing instructions after the agent followed unwanted text embedded in job descriptions.
The reported outcome is repeated applications with a tracking log, not interviews or a job offer. Example rows and broad timings illustrate the system but are not an independently audited success count. The reusable point is matching documents and form actions to a reviewed job list, with barriers and skipped work visible.
Prompt
Original prompt
Source: Blog · Artur Erofeev
Frequently asked questions
Who chooses which jobs are applied to?
The agent collects matches, but the author reviews the CSV and removes unsuitable entries before proceeding.
Does this establish hiring success?
No interviews or job offers are reported as outcomes of this workflow.
What happens at authentication barriers?
CAPTCHAs are recorded as blocked, and two-factor steps require the human to intervene.
More details
Remaining steps
- Log submissions and skipped barriers and request human help where needed.
About this account
Claude Code searches and deduplicates jobs, writes cover letters and Selenium scripts, submits forms and logs outcomes.
Firsthand primary account opened and read; original publication date checked in primary metadata or visible timestamp. Outcomes are author-reported, not independently reproduced. Firsthand guide and logs, no independently audited application count or hiring result. Original prompt excerpt reproduced exactly from the source; excerpt only, not the complete prompt.
Credit: Artur Erofeev
Source published 17 Mar 2026
Source checked 15 Oct 2026
Tools and topics
Claude Code · Chrome · Selenium · CSV logs · Markdown skill
Work · Reported deployment
Editorial completeness: 100/100. Measures documented fields, not agent performance or verified outcomes.