All agents

HR Recruiter Agent

Turns a single role spec into a job description, a screening rubric, and personalized candidate outreach in one sub-minute run.

View in marketplace Open dashboard

The HR Recruiter Agent automates the top of your hiring funnel. Give it a role to hire and a location, and it writes a crisp job description, builds an ideal-candidate screening rubric, and drafts a personalized outreach message — optionally pulling live market and salary context from the web. If a candidate email is provided and you've connected a mailbox, it can send the outreach for you instead of just drafting it.

What it does

The agent acts as a technical recruiter that handles the repetitive, time-consuming part of hiring: defining the role, deciding what "good" looks like, and reaching out. On each run it reads the role and location from your deployment config, optionally researches current market and salary context via web search and web fetch, then produces three artifacts together — a job description, a screening rubric (must-haves, nice-to-haves, red flags), and a tailored outreach message. All three are written and returned in a single recorded output, so you spend your time only on the shortlist instead of on blank-page busywork.

Operationally it runs as a deployed agent on AgentOS with a capped 5-turn budget, so runs complete quickly (typically under a minute). It has exactly four tools available — web_search and web_fetch for optional research, send_email for delivering outreach, and record_output to package the final deliverable. Sending only happens when a candidate email is supplied; otherwise everything stays as a draft for you to review and send yourself. Note: this agent has no interview-scheduling or calendar capability and no ATS/sourcing-platform integrations — it produces the JD, rubric, and outreach text only.

How it works

Your Business
1
Read the role
Ingests the configured 'Role to hire' (title + key requirements) and 'Location / remote' fields from the deployment and frames the run: source and screen for this role.
2
Research market context (optional)
When useful, uses web_search to find current market/salary signals and web_fetch to read a specific page, grounding the JD and outreach in real data rather than guesswork.
3
Write the job description
Drafts a crisp, role-specific JD reflecting the title, key requirements, and location/remote setup.
4
Build the screening rubric
Produces an ideal-candidate rubric — must-haves, nice-to-haves, and red flags — so every candidate is screened consistently against the same bar.
5
Draft personalized outreach
Writes a tailored outreach message for a shortlisted candidate, referencing specifics rather than a generic template.
6
Send or draft
If a candidate email is provided, sends the outreach via send_email; otherwise leaves it as a draft for you to send.
7
Record the deliverable
Calls record_output exactly once with all three artifacts (JD + rubric + outreach), which appears in the deployment's run log.
Outcomes delivered

Setting it up — owner / admin

  1. 1
    Deploy from the marketplace
    Subscribe to the HR Recruiter Agent, then open its deployment at /dashboard/agents/[id]. The page shows the deployment status, a 'Run now' button, a configuration panel, and a run log.
  2. 2
    Connect a mailbox (optional, for sending)
    Connect a mailbox so the agent can actually send outreach via send_email. Without a connection — or without a candidate email — it still drafts the outreach for you to send yourself.
  3. 3
    Fill the configuration fields
    In the config panel set 'Role to hire' (the textarea — title plus key requirements, e.g. 'Senior Full-Stack Engineer (TypeScript, Next.js)') and 'Location / remote' (e.g. 'Remote, EU timezones'). Save the configuration.
  4. 4
    Do the first run
    Click 'Run now'. Within about a minute the JD, rubric, and outreach draft appear in the run log below. There are no required external tools — the research tools (web_search, web_fetch) are optional.
  5. 5
    Set up triggering
    This agent has no built-in schedule field, so it's on-demand by default — run it per new req from the dashboard. For event-driven kickoff (e.g. when a new role opens in your ATS), POST to the deployment's per-deployment webhook (/api/hooks/[token]) to trigger a run automatically.
  6. 6
    Review and refine
    Read the recorded output, tweak the role spec to reshape the rubric and tone, and re-run. The rubric is driven by the role spec, so refining the field is how you customize screening per deployment.

Using it day to day — your team

  1. 1
    Open a new requisition
    When a role opens, update the 'Role to hire' and 'Location / remote' fields on the deployment to match the new req, then save.
  2. 2
    Run it
    Hit 'Run now' on /dashboard/agents/[id] (or let an ATS/webhook trigger fire). Each run yields a JD, a screening rubric, and an outreach draft for the role.
  3. 3
    Review the deliverable
    Read the recorded output in the run log: the job description to post, the must-have/nice-to-have/red-flag rubric to screen against, and the personalized outreach message.
  4. 4
    Send outreach
    If you supplied a candidate email and connected a mailbox, the agent already sent the message — confirm in your Sent folder. Otherwise copy the draft, personalize the final details, and send it from your inbox.
  5. 5
    Screen candidates consistently
    Apply the same rubric to every applicant for the role so screening stays consistent, and only escalate the shortlist for interviews.

Use cases

New req kickoff
Turn a freshly approved role into a ready-to-post JD, a screening rubric, and outreach in a single run — no blank-page time.
Evergreen pipeline building
Keep outreach flowing for always-open roles by re-running the agent and feeding fresh candidates through the same rubric.
Consistent screening
Apply one rubric (must-haves, nice-to-haves, red flags) to every candidate so evaluations stay objective and comparable.
Market-aware hiring
Let the agent pull current market and salary context via web research so the JD and outreach reflect real conditions, not assumptions.

What to expect

  • A crisp, role-specific job description per run
  • An ideal-candidate screening rubric: must-haves, nice-to-haves, and red flags
  • A personalized outreach message — drafted, or sent via send_email when a candidate email is provided and a mailbox is connected
  • A faster, more consistent top-of-funnel with less recruiter busywork
  • All three artifacts recorded together in the deployment run log for review and reuse

Metrics to watch

  • Runs complete in under a minute (5-turn cap keeps it fast)
  • Every run returns all three artifacts via a single record_output call
  • Outreach reply rate on sent messages (a proxy for personalization quality)
  • Shortlist quality / pass-through rate when the rubric is applied consistently
  • Time-to-first-outreach per new requisition

FAQ

Does it contact candidates directly?
It drafts outreach by default. It only sends if you provide a candidate email and have a mailbox connected; then it sends via the send_email tool. With no candidate email (or no mailbox), it simply leaves the outreach as a draft for you to send.
Can I customize the screening rubric?
Yes. The rubric is generated from the 'Role to hire' spec, so refining that field — adding requirements, seniority, or constraints — reshapes the must-haves, nice-to-haves, and red flags. You can tune it per deployment.
Do I need to connect LinkedIn, Indeed, or an ATS?
No. The agent's only tools are web_search and web_fetch (optional research), send_email (optional sending), and record_output. Sourcing platforms and ATS systems are not wired in as tools; you feed it the role and review the drafted artifacts.
Can it schedule interviews or sync to my calendar?
No. This agent has no calendar or scheduling tool. It produces the job description, screening rubric, and outreach message only — booking interviews is something you do yourself outside the agent.
Can it run automatically when a role opens?
It's on-demand by default (no schedule field), so you typically click 'Run now' per req. For automated kickoff you can trigger a run by POSTing to the deployment's per-deployment webhook token at /api/hooks/[token], e.g. from your ATS when a new role is created.