Skip to content

Repository files navigation

career-ops License Platforms Modules

CAREER-OPS

NO SLOP. JUST STRATEGY.

An open-source AI career coaching skill for Claude Code, Codex, Gemini CLI, Cursor, and more. Resume optimization, interview prep, job search strategy, and anti-slop quality gates — backed by peer-reviewed research.

View on GitHub Visit Website Changelog

career-ops four-stage engine: interviewer mode, smart routing, anti-slop gate, and session memory — brutalist UI on blueprint grid


── FOUR-STAGE ENGINE

Every session follows the same pipeline: interview first → route to the right module → score the output → learn for next time.

# STAGE DESCRIPTION TAGS
01 INTERVIEWER MODE Motivational Interviewing framework that mines accomplishments through structured questioning — not self-assessment. MI Framework · 5 Phases
02 SMART ROUTING Your request is matched to the right reference module. "Tailor my resume" auto-loads resume.md. 10 Modules · Auto-Load
03 ANTI-SLOP GATE Output scored on 5 dimensions before delivery. Below 35/50, it gets rewritten automatically. 35/50 Min · 5 Dimensions
04 SESSION MEMORY Profile updates after every session — skills discovered, achievements surfaced, preferences saved. Persistent · Profile

── INSTALL IN 30 SECONDS

Start a session and say "Help me with my resume." The skill activates automatically on any career-related prompt.

git clone https://github.com/poferraz/career-ops.git ~/.claude/skills/career-ops

No API keys required. No MCP servers. No cloud services. Everything runs locally in your agent's context.

PLATFORM METHOD STATUS
Claude Code git clone into ~/.claude/skills/ ✓ Native
Codex bash scripts/convert.sh codex ✓ Supported
Gemini CLI bash scripts/convert.sh gemini ✓ Supported
Cursor bash scripts/convert.sh cursor ✓ Supported
Antigravity bash scripts/convert.sh antigravity ✓ Supported
Windsurf bash scripts/convert.sh windsurf ✓ Supported
OpenCode bash scripts/convert.sh opencode ✓ Supported

Other platforms: The convert script outputs platform-specific formats. See integrations.md for setup details.


── ANTI-SLOP QUALITY GATE

Every response is scored on 5 dimensions before you see it. Below 35/50, it gets rewritten automatically. No exceptions.

DIMENSION SCORE WHAT IT CATCHES
Directness 9/10 Hedging, filler, passive voice
Rhythm 8/10 Monotonous sentence structure
Trust 9/10 Unsupported claims, invented statistics
Authenticity 8/10 Chatbot tone, buzzword density
Density 8/10 Sentences that earn no place
TOTAL 42/50 Minimum passing threshold: 35
BEFORE AFTER
Resume bullet "Responsible for managing team projects and improving efficiency" "Led 4-person team delivering $2M platform migration 3 weeks early"
Interview answer "I'm a great communicator and team player" "Cut stakeholder meeting time 40% by switching to async Loom updates"
LinkedIn headline "Passionate marketing professional driving brand growth" "Grew SEM revenue $1.4M in 11 months — B2B SaaS, Series A"

── REFERENCE MODULES

Ten evidence-backed modules covering every stage of the multi-agent career coaching workflow.

MODULE COVERS
resume.md Eye-tracking data, ATS mechanics, achievement reframing via RISE method
cover-letter.md Gap framing, AI detection reality, Hook → Match → Bridge → Ask structure
linkedin.md Profile optimization, headline strategy, posting cadence, engagement hierarchy
interview.md STAR validation, predicted questions from JD, negotiation frameworks (ZOPA/BATNA)
outreach.md Connection request acceptance rates, cold message frameworks, InMail vs email
job-search.md Channel effectiveness data, diminishing returns curve, industry-specific duration
career-strategy.md GROW/OSCAR/CLEAR coaching frameworks, cross-industry mobility mapping
market-data.md Hiring funnel conversion rates, salary research sources, remote work snapshot
anti-slop.md Scoring rubric, banned phrase list, familiarity problem detection
research-scout.md Live web search protocol, recency checks, mandatory [UNVERIFIED] tagging

── RESEARCH-BACKED

Every claim is tagged T1–T4. No invented statistics. No plausible-sounding numbers. T4 claims are excluded entirely.

TIER LABEL SOURCE TYPE
T1 Verified Peer-reviewed studies, RCTs, meta-analyses
T2 Reported Controlled callback experiments, platform studies
T3 Expert Conventional wisdom from recognized practitioners
T4 Excluded Contradicted by evidence, origin unknown

Built through 5 rounds of deep research with goal-blind prompting and contradiction checking. See research-methodology.md for full methodology.


── CONTRIBUTE

career-ops runs on evidence, not vibes. Every contribution must clear the same anti-slop bar as the skill itself.

Contributing Changelog Security Code of Conduct Support

MIT License — open for all platforms and all workers.

About

An agent skill for Claude Code, Codex, Gemini CLI, Cursor, and more.

Topics

Resources

Code of conduct

Contributing

Security policy

Stars

6 stars

Watchers

1 watching

Forks

Releases

Packages

Used by

Contributors

Languages