---
name: career-ops
description: AI job search command center — evaluate offers, generate CVs, scan portals, track applications. Adapted from Claude Code to Hermes Agent.
user_invocable: true
args: mode
---

# career-ops — Router (Hermes Edition)

Adapted from Claude Code's career-ops system to work with Hermes Agent's tooling.

## Mode Routing

When invoked (via `/skill career-ops` or as preloaded skill), determine the mode from user input:

| Input | Mode |
|-------|------|
| (empty / no args) | `discovery` — Show command menu |
| JD text or URL (no sub-command) | **`auto-pipeline`** |
| `oferta` | `oferta` |
| `ofertas` | `ofertas` |
| `contacto` | `contacto` |
| `deep` | `deep` |
| `pdf` | `pdf` |
| `training` | `training` |
| `project` | `project` |
| `tracker` | `tracker` |
| `pipeline` | `pipeline` |
| `apply` | `apply` |
| `scan` | `scan` |
| `batch` | `batch` |

**Auto-pipeline detection:** If input is not a known sub-command AND contains JD text (keywords: "responsibilities", "requirements", "qualifications", "about the role", "we're looking for", company name + role) or a URL to a JD, execute `auto-pipeline`.

## Discovery Mode (no arguments)

Show this menu:

```
career-ops — Command Center (Hermes Edition)

Available commands:
  career-ops {JD or URL}   → AUTO-PIPELINE: evaluate + report + PDF + tracker (paste text or URL)
  career-ops pipeline      → Process pending URLs from inbox (data/pipeline.md)
  career-ops oferta        → Evaluation only A-F (no auto PDF)
  career-ops ofertas       → Compare and rank multiple offers
  career-ops contacto      → LinkedIn power move: find contacts + draft message
  career-ops deep          → Deep research prompt about company
  career-ops pdf           → Generate ATS-optimized CV PDF
  career-ops training      → Evaluate training/certification courses
  career-ops project       → Evaluate portfolio project ideas
  career-ops apply         → Live application form assistant
  career-ops tracker       → View application tracker
  career-ops scan          → Scan configured company portals for new listings
  career-ops batch         → Batch process multiple URLs
```

## Execution Rules

1. **Read profile first:** Always load `config/profile.yml` for candidate identity.
2. **Read CV:** Always read `cv.md` (project root or HERMES_HOME/career-ops/) for experience data.
3. **JD extraction:** For URLs, use `browser_navigate` + `browser_snapshot` (SPAs) → `web_extract` (static) → `web_search` (fallback).
4. **Reports go to:** `reports/{###}-{company-slug}-{YYYY-MM-DD}.md`
5. **Tracker is at:** `data/applications.md`
6. **Pipeline inbox is at:** `data/pipeline.md`

## Sources of Truth

| File | Path | Purpose |
|------|------|---------|
| SKILL.md | This file | Router |
| modes/ | `modes/*.md` | Individual mode definitions |
| profile.yml | `config/profile.yml` | Candidate identity and targets |
| cv.md | project root or skill dir | CV in markdown |
| states.yml | `templates/states.yml` | Canonical application states |
| portals.yml | project root | Portal scanner config |

## Scan Mode Implementation Details

When processing `career-ops scan`:

1. **Load config:** Read `portals.yml` for tracked companies + search queries
2. **WebSearch sweep:** Run 20+ queries across site:jobs.ashbyhq.com, site:job-boards.greenhouse.io, site:jobs.lever.co, and general ML role queries
3. **Filter results:** Match against candidate's profile (roles, tech stack, location constraints) — skip PM/sales/junior roles
4. **Rank tiers:** A-tier (strong match), B-tier (potential), Skip (wrong fit)

**CRITICAL ASHBY PATTERN:**
- Individual Ashby job URLs from web_search results (jobs.ashbyhq.com/{company}/{uuid}) are FREQUENTLY stale/dead with 404 errors
- ALWAYS navigate to the company page first: `jobs.ashbyhq.com/{slug}` (found in portals.yml `careers_url`)
- Use `browser_navigate` + `browser_snapshot(full=true)` — web_extract returns near-empty on Ashby (JS-rendered SPA)
- From company page, use snapshot to find current open roles with their fresh URLs, then click into each for JD