skillich

---

What we found

Finding	Details
Skilled Trades are the most AI-resilient function	Trades average just 21% automation risk -- the lowest of all 17 functions. You can't automate pulling wire through walls or soldering pipes.
8 roles have "low" AI impact	4 of them are skilled trades (electrician, plumber, HVAC, automotive). Physical dexterity and on-site judgment can't be automated.
Marketing is the most exposed function	54% average automation risk across all marketing skills. Content generation, ad optimization, and analytics are rapidly AI-driven.
7 roles are rated "very-high" AI impact	Including Content Writer, Paralegal, Medical Coder, Technical Writer, and SDR. These roles are transforming fastest.
The most resilient skills?	Electrical Installation, Live Performance, Apprentice Mentoring, and Conflict Resolution -- all at 5%. Hands-on craft and human connection can't be delegated to a model.

---

Top 15 Skills for Every AI User

No matter your role, these are the skills that matter most in the agentic age:

Tier	Skills	Why
AI Fluency	Prompt Design, AI Output Evaluation, Model Behavior Understanding	Direct AI effectively and spot when it's wrong
Human Anchors	Cross-Functional Collaboration, Stakeholder Management, Mentoring, Ethical Judgment, Conflict Resolution	What AI can't do -- your career insurance
AI-Optimised Skills	Data-Driven Decisions, Strategic Communication, Process Improvement, Statistical Analysis, Change Management	Where AI amplifies you 10x
Domain Depth	Your core expertise, Adaptability	What makes AI work for you specifically

Read the full guide → | Skills by function →

---

Try It Now -- No Install Required

Get My Action Plan

1. Type your job title -- accountant, PM, backend engineer, electrician, paralegal. Acronyms work too (pm, sre, sdr, swe).
2. Your plan appears, with a level switcher on it -- flip between junior and staff to see how the advice changes.

You get every skill in your role sorted into three actions, plus what "good" looks like at your next level and a concrete way to prove you got there.

60 seconds. No signup. No cost. Nothing leaves your browser.

And you can keep it:

• Copy link -- every plan has its own URL (.../#/plan/product/product_manager/senior). Send it to your manager.
• Download as Markdown -- drop it into Notion, Obsidian, or your 1:1 doc.
• Print / save as PDF -- a real print stylesheet, not a screenshot.

Prefer the terminal? skillich plan "product manager" --save plan.md gives you the same thing.

---

What's Inside

For every role across 17 functions -- engineering, product, design, data, marketing, operations, leadership, finance, legal, sales, support, HR, healthcare, AI & agents, creative, research, and skilled trades:

• What AI is automating -- stop doing these manually
• What AI is augmenting -- learn to use AI as your 10x multiplier here
• What remains human -- pour your career investment here
• Proficiency levels -- what "good" looks like from junior through principal
• AI impact scores -- every skill rated 0.0 (fully human) to 1.0 (fully automated)
• Career transitions -- which roles you can move to based on skill overlap
• Self-assessment -- score yourself across 15 AI-era skills and get personalised gap analysis
• Learning resources -- curated courses, books, and free guides for every skill area

Who This Is For

Wondering how AI changes your job? Get your action plan. 60 seconds.

Building a team development plan? skillich export --function engineering --format csv gives you a spreadsheet with every skill, rating, and proficiency level.

Designing career frameworks? The taxonomy has 5 proficiency levels per skill with observable behaviors. Use it as your leveling foundation.

Building AI-powered career tools? Machine-readable YAML with Python SDK, MCP server, and OpenAI/Anthropic tool export.

---

This Project Needs You

skillich was built from research and industry frameworks. The ratings become trustworthy when real practitioners validate them.

If you're an accountant, recruiter, PM, engineer, or anything in between -- your experience is the most valuable thing missing:

Submit feedback -- 2 minutes, no coding. Tell us what matches your experience and what doesn't.

Propose a rating change -- share evidence from your work:

"I've been a Controller for 8 years. The 'Month-End Close' rating of 0.4 is too high -- AI generates journal entries but someone still investigates every variance. I'd say 0.25."

Share your playbook story -- tried the 6-step playbook? Tell us what happened after 90 days.

Star this repo to get notified when ratings are updated for your function.

---

Coverage

Function	Roles	Avg AI Impact	Most Resilient Skill
AI & Agents	5	31%	AI Ethics & Responsible Deployment
Creative & Media	5	32%	Live Performance
Research & Academia	5	29%	Mentoring & Supervision
Skilled Trades	5	21%	Electrical Installation
Engineering	10	46%	Product & Design Collaboration
Product	4	42%	Stakeholder Management
Design	4	41%	Cross-Functional Collaboration
Data & Analytics	4	49%	Stakeholder Communication
Marketing & Growth	5	54%	Brand Strategy
Operations	5	41%	Servant Leadership
Leadership	6	33%	Engineering Culture
Finance	5	46%	Team Management
Legal	5	47%	Cross-Functional Collaboration
Sales	5	52%	Relationship Building
Customer Support	5	46%	Empathy & Emotional Intelligence
HR & People Ops	5	46%	Conflict Resolution
Healthcare	5	42%	Cross-Functional Collaboration

---

CLI

pip install "skillich[cli]"

skillich plan "product manager"                          # Your action plan, start here
skillich plan pm --level senior --save plan.md           # Same plan, saved as Markdown
skillich plan electrician --transitions                  # ...plus where your skills transfer
skillich plan pm --format json                           # Machine-readable

skillich explore                                         # Browse all 17 functions
skillich explore --function engineering --role backend   # Full role detail
skillich impact --resilient --limit 10                   # Most AI-resilient skills
skillich impact --function engineering --compare         # Compare roles in a function
skillich impact --resilient --format json                # Machine-readable output
skillich learn --function engineering --role backend --level senior  # Learning path
skillich transitions --function engineering --role backend  # Career transitions
skillich narrate --function engineering --role backend   # AI-written career insight (Ollama)
skillich audit --stale-after 180                         # Find stale taxonomy data
skillich export --function engineering --format csv      # Export for spreadsheets

MCP Server (Claude Desktop / Claude Code)

pip install "skillich[mcp]"

Until the first PyPI release (tag v0.2.0), install from source:

pip install "skillich[mcp] @ git+https://github.com/salahuddinuqaili/skillich.git"

Add to Claude Desktop config:

{
  "mcpServers": {
    "skillich": {
      "command": "skillich-mcp"
    }
  }
}

Then ask Claude: "What skills should a mid-level PM invest in given AI?"

Claude answers from the action_plan tool -- one call takes a plain job title and returns the full invest / partner / delegate breakdown, the next-level gap, and a shareable link. Ambiguous titles ("engineer", "analyst") come back as a list of candidates rather than a confident guess.

Tool	What it does
action_plan	Start here. Job title in, full action plan out
explore_roles	Browse functions, roles, skills, proficiency levels
analyze_ai_impact	Compare AI impact across roles and skills
assess_skills	Gap analysis against a target role
generate_learning_path	Prioritised development plan
advise_career	Skill transfer and transition analysis

Python API

from core.taxonomy import TaxonomyLoader

loader = TaxonomyLoader("taxonomy")
role = loader.get_role("engineering", "backend")
print(role.ai_impact.summary)

# Most AI-resilient skills
resilient = loader.get_skills_by_ai_impact(max_rating=0.3)

from core.plan import build_plan, find_roles

find_roles(loader, "pm")[0]["role_name"]          # 'Product Manager'

plan = build_plan(loader, "product", "product_manager", "senior")
plan.exposure                                      # 0.33
[s.name for s in plan.invest]                      # skills AI can't touch
print(plan.to_markdown())                          # the document the web app downloads

core/plan.py holds the action-band thresholds used by the CLI, the MCP server and the web app -- change them there and all three follow.

Agent Tool Export

from core.registry import SkillRegistry
registry = SkillRegistry()
registry.auto_discover("skills")

openai_tools = registry.to_openai_tools()    # OpenAI function calling
anthropic_tools = registry.to_anthropic_tools()  # Anthropic tool_use
mcp_tools = registry.to_mcp_tools()          # Model Context Protocol

Built-in Agent Skills

Skill	What It Does
role_explorer	Browse functions, roles, skills, and proficiency levels
ai_impact_analyzer	Compare AI impact across roles, find most/least automatable skills
skill_assessor	Assess your skills against a target role, get gap analysis
learning_path	Generate a prioritized development plan weighted by AI resilience
career_advisor	Cross-role skill transfer analysis, career transition recommendations

---

Methodology

AI impact ratings (0.0-1.0) are expert estimates informed by the Anthropic Economic Index, GitHub Copilot research, ILO automation studies, and industry career frameworks. These ratings have not yet been validated by practitioners -- that's why your feedback matters.

Every role has a confidence level (low/medium/high) and a last_reviewed date. Most are currently medium. The path to high is practitioner validation.

Known limitations: Ratings are ranges, not precise numbers. Context matters (a 0.5 at a legacy enterprise may be 0.7 at an AI-native startup). Ratings decay as AI capabilities improve — which is why we follow a quarterly review cadence. See the full methodology docs.

Contributing

The most impactful contributions don't require code:

1. Validate your role -- do the ratings match your experience?
2. Share your story -- followed the playbook? Tell us what happened
3. Propose changes -- with evidence from your work
4. Add new roles -- see CONTRIBUTING.md

License

MIT