Hyreflow

Quickstart

From zero to a qualified shortlist in under a minute.

Got an open role? Hyreflow takes it from a job description to a shortlist of qualified candidates with reachable contacts — driven by your agent of choice over a hosted, credit-metered backend.

Install

One command installs the CLI, the agent skills, and connects your account in the browser (no token paste — +25 free credits).

curl -fsSL https://recruit.hyreflow.ai/api/v2/cli/install | bash

This installs the hyreflow CLI to ~/.local/bin, unpacks the agent skills to ~/.agents/skills, and links them into every supported agent's skills directory found on the machine — then prints a few starter recipes to run.

On Windows, use the npm tab instead — there's no PowerShell equivalent of this script.

The CLI is on public npm. Installing it lands the binary; hyreflow setup adds the agent skills and your account.

npm install -g hyreflow
hyreflow setup

hyreflow setup installs the agent skills, opens your browser to sign in, and confirms the workspace. It's resumable — if you haven't approved yet it says so, and re-running continues the same sign-in.

Blocked from npmjs.com? Add --registry https://recruit.hyreflow.ai/api/v2/npm/. Want the binary only (CI, image builds)? Set HYREFLOW_NO_SKILLS=1.

On Windows, npm puts the shim in %APPDATA%\npm, which a shell opened before the install doesn't have on its PATH. If hyreflow isn't recognized, pick it up in the current shell with $env:Path = "$env:APPDATA\npm;$env:Path", or open a new terminal.

Install the CLI, then authenticate with a token from the dashboard.

curl -fsSL https://recruit.hyreflow.ai/api/v2/cli/install | bash
hyreflow auth login --token hf_live_xxx
hyreflow skills install

On Windows, install with npm install -g hyreflow first, then run the same auth login / skills install commands.

Prefer a project-local install over a global one? npm install -g --prefix "<project>/.hyreflow/runtime" hyreflow puts the binary under the project instead of your global npm prefix — invoke it by full path (POSIX: <project>/.hyreflow/runtime/bin/hyreflow; Windows: <project>\.hyreflow\runtime\hyreflow.cmd, since npm places shims at the prefix root there, not under bin\). Skills and config still install to your home directory either way.

Getting started

Install the harness

Run the install command above. It connects your account and seeds your credit balance.

hyreflow billing balance     # confirm your +25 starting credits

Load the skill

In your agent, invoke the meta skill with your task in plain English.

/hyreflow-recruit Source 20 senior backend engineers in Berlin with verified personal emails

The agent reads the matching recipe, publishes a live plan to the Playground, then runs the pipeline.

Run a recipe

Or start from a guided demo:

/hyreflow-quickstart

What just happened

The agent didn't call one vendor — it ran a waterfall. For sourcing it searched the people-DBs by title + location; for contact it ran the personal_email waterfall (see Integrations Overview for the provider order), stopping at the first usable hit.

Waterfalls stop at the first hit and charge only on a usable result — soft misses are free. Ordering providers cheap → expensive means you typically pay one provider, not all of them, and recover 20–40% more contacts than any single source.

Drive the API directly

Prefer the CLI? Every step is a metered HTTP call.

hyreflow tools execute people_search \
  --payload '{"titles":["Backend Engineer"],"locations":["Berlin"],"skills":["Go"],"limit":20}'
hyreflow enrich --input candidates.csv --output enriched.csv \
  --with '{"alias":"email","tool":"personal_email","payload":{"linkedin_url":"{{linkedin_url}}"}}'
hyreflow qualify --job @jd.txt --candidates @candidates.json
hyreflow billing balance

Scaling up

Running a pilot first is the discipline that keeps credit spend honest.

Advanced CLI: pilot before the full pull

Always preview a slice before a bulk run:

# 1) preview the call shape and cost without charging
hyreflow tools execute people_search --dry-run --payload '{"titles":["CTO"],"limit":1}'

# 2) pilot a single row (--rows is a row RANGE, end-exclusive) through the enrich waterfall
hyreflow enrich --input all_candidates.csv --output enriched.csv --rows 0:1 \
  --with '{"alias":"email","tool":"personal_email","payload":{"linkedin_url":"{{linkedin_url}}"}}'

# 3) approve, then run the full pull (drop --rows, or widen the range)
hyreflow enrich --input all_candidates.csv --output enriched.csv \
  --with '{"alias":"email","tool":"personal_email","payload":{"linkedin_url":"{{linkedin_url}}"}}'

Use --dry-run on any tool call to see the resolved provider, params, and credit cost before committing.

Next steps

On this page