Skip to content
Conduit

skill

hermes-merchant

A collection of portable agent skills that scrape ML/AI jobs, score them against your profile, and auto-fill Greenhouse applications.

arimanyus/hermes-merchantcommunitylistedunreviewednot running here yet
Use skill

Set up

Skills are installed by placing their folder where your agent loads skills.

Install instructions come from the README

Skills do not use an MCP config. Follow the steps in the README below; one-click install arrives with accounts.

README

synced Oct 14, 2026

hermes-merchant

A collection of portable agent skills for running an automated job hunt — scraping ML/AI roles across Greenhouse, AshbyHQ, and Lever; scoring them against your profile; and filling Greenhouse applications via browser automation.

Compatible with Hermes, Claude Code, OpenClaw, and any other SKILL.md-compatible agent.

Merchant skills are a class of skills that help you sell yourself — to an ATS, a recruiter, a platform. Start here: the job-hunt set. Expect more merchant skills over time.

Skills in this repo

SkillWhat it does
job-scraper-pipelineScrape ML/AI jobs from Greenhouse, AshbyHQ, Lever. Dedupe, score 0–100, classify tier-1/2/3. Encoded field-tested notes on which sources work and which are blocked.
greenhouse-job-applicationFill a Greenhouse application in one JS expression using the React native-value setter. Verify fields, detect reCAPTCHA, fall back to careers@ email when React dropdowns can't be driven.
browser-harness-ats-automationCDP + cross-origin iframe patterns for Ashby/Greenhouse/Workday: attach to the form iframe, set file inputs, click hidden work-authorization checkboxes, intercept the GraphQL submit response.

Install

All three skills live as folders with a SKILL.md at the top. Pick whichever install path matches your agent.

Hermes

git clone https://github.com/arimanyus/hermes-merchant ~/hermes-merchant

# Symlink the skills into your Hermes skills directory
ln -s ~/hermes-merchant/job-scraper-pipeline            ~/.hermes/skills/productivity/job-scraper-pipeline
ln -s ~/hermes-merchant/greenhouse-job-application      ~/.hermes/skills/productivity/greenhouse-job-application
ln -s ~/hermes-merchant/browser-harness-ats-automation  ~/.hermes/skills/productivity/browser-harness-ats-automation

Then hermes skills list should show all three under productivity.

Claude Code

git clone https://github.com/arimanyus/hermes-merchant ~/.claude/skills/hermes-merchant

Each subfolder is a self-contained skill. Reference them from your CLAUDE.md:

- Job scraping    → read `~/.claude/skills/hermes-merchant/job-scraper-pipeline/SKILL.md`
- Apply on Greenhouse → read `~/.claude/skills/hermes-merchant/greenhouse-job-application/SKILL.md`
- ATS browser automation (Ashby, Greenhouse) → read `~/.claude/skills/hermes-merchant/browser-harness-ats-automation/SKILL.md`

OpenClaw

git clone https://github.com/arimanyus/hermes-merchant ~/.openclaw/skills/hermes-merchant

Configure your profile

All three skills read from a single profile.yaml at the repo root. This file is gitignored so your PII never leaves your machine.

cd ~/hermes-merchant
cp profile.yaml.example profile.yaml
$EDITOR profile.yaml

At a minimum, set:

  • identity.first_name, last_name, email, phone, country
  • resume.path — absolute path to your resume PDF
  • work_authorization — us_authorized and needs_visa_sponsorship
  • scoring.required_keywords — tune to your target roles

Everything else has sensible defaults.

What the pipeline looks like end-to-end

┌───────────────────────────┐
│  job-scraper-pipeline     │  every 15 min
│  Greenhouse / Ashby /     │──┐
│  Lever → score → queue    │  │
└───────────────────────────┘  │
                               ▼
                   ┌───────────────────────────┐
                   │  state/queue/jobs.json    │
                   │  [{tier-1, tier-2, ...}]  │
                   └───────────────────────────┘
                               │
                               ▼
┌──────────────────────────────────────────────────┐
│  greenhouse-job-application /                    │  every 30 min
│  browser-harness-ats-automation                  │
│  → fill form, attach resume, submit or email     │
└──────────────────────────────────────────────────┘
                               │
                               ▼
                   ┌───────────────────────────┐
                   │  state/applied/           │
                   │  applied.json             │
                   └───────────────────────────┘

Why three skills instead of one

Each skill is short, scoped to one kind of failure, and can be read by the agent on-demand. When Greenhouse ships a new React widget and something breaks, you only have to touch one file — the others stay stable. It also lets you cherry-pick; if you just want the scraper and plan to apply manually, skip the two applier skills.

Contributing

PRs welcome, especially:

  • New ATS platforms (Lever, Workday, SmartRecruiters application filling)
  • More careers@ fallbacks in profile.yaml.example
  • Better scoring heuristics for specific role families

Please keep PII out of your diffs — if you catch a hardcoded name/email in a PR, call it out.

License

MIT — see LICENSE.

↑↓ navigate openesc close