# Ai Career Threat Index > Open dataset: 300 US professions scored on AI displacement risk — four published sub-scores per role incl. exposure to autonomous agents. Quarterly updates, SOC crosswalk, interactive explorer, MCP server. MIT. - **Type:** MCP server - **Install:** `agentstack add mcp-jott2121-ai-career-threat-index` - **Verified:** Yes — security-reviewed for prompt injection and unsafe behavior - **Seller:** [Jott2121](https://agentstack.voostack.com/s/jott2121) - **Installs:** 0 - **Category:** [Integrations](https://agentstack.voostack.com/c/integrations) - **Latest version:** 0.1.0 - **License:** MIT - **Upstream author:** [Jott2121](https://github.com/Jott2121) - **Source:** https://github.com/Jott2121/ai-career-threat-index - **Website:** https://jott2121.github.io/ai-career-threat-index/ ## Install ```sh agentstack add mcp-jott2121-ai-career-threat-index ``` Requires the [AgentStack CLI](https://agentstack.voostack.com/docs/cli). Works with Claude Code, Cursor, and any MCP-compatible agent. ## About # AI Career Threat Index > How exposed is your job to AI? 300 professions scored on an open rubric — > four published sub-scores per role, including exposure to **autonomous agents**. > Updated quarterly. MIT licensed. [](LICENSE) [](https://github.com/Jott2121/ai-career-threat-index/actions/workflows/validate.yml) [](https://huggingface.co/datasets/Jott2121/ai-career-threat-index) [](https://doi.org/10.5281/zenodo.21227026) **[▶ Explore all 300 roles interactively](https://jott2121.github.io/ai-career-threat-index/)** · [Winners & Losers this quarter](reports/2026-q3-winners-losers.md) · [Methodology](pipeline/RUBRIC.md) · [Ask it through your AI assistant (MCP)](mcp/) ## What this is The AI Career Threat Index scores **300 US professions** on AI displacement risk, 0–100. Unlike a single black-box number, every score decomposes into **four published sub-scores** with written rationales: | Sub-score | Question it answers | |---|---| | `taskAutomation` | What share of the role's core work could AI do end-to-end today, at ≥90% reliability? | | `toolMaturity` | How mature and deployed are the AI tools targeting those tasks? | | `adoption` | What share of employers actually use AI for those tasks in production? | | `agenticExposure` | What share of the role could **autonomous multi-step agents** own — not copilots, agents? | combined by an open formula (task automation sets the ceiling; maturity and adoption determine how much is realized; agentic exposure adds forward pressure): ``` score = taskAutomation × (0.45 + 0.30·toolMaturity/100 + 0.25·adoption/100) + 0.10·agenticExposure ``` The `agenticExposure` factor is, to our knowledge, the first published per-occupation scoring of exposure to autonomous agents specifically — the thing that changed between 2024's copilots and 2026's agent deployments. **Honesty note:** scores are *structured editorial estimates* made against a [published rubric](pipeline/RUBRIC.md) with named anchors and calibration examples — informed by O*NET task lists, BLS OES data, and public adoption research. They are not measurements, and we don't claim otherwise. Every input is in this repo; if you disagree with a sub-score, [open an issue](../../issues) and argue with the rationale — that's the point of publishing them. ## Get the data ```bash git clone https://github.com/Jott2121/ai-career-threat-index.git # or grab the files directly: curl -LO https://raw.githubusercontent.com/Jott2121/ai-career-threat-index/main/data/ai-career-threat-index.json curl -LO https://raw.githubusercontent.com/Jott2121/ai-career-threat-index/main/data/ai-career-threat-index.csv curl -LO https://raw.githubusercontent.com/Jott2121/ai-career-threat-index/main/data/soc-crosswalk.csv ``` | File | What's in it | |---|---| | [`data/ai-career-threat-index.json`](data/ai-career-threat-index.json) | Full dataset: 300 roles, sub-scores, rationales, tasks, defense skills, quarterly history | | [`data/ai-career-threat-index.csv`](data/ai-career-threat-index.csv) | Flat table, one row per role — Excel/pandas/R ready | | [`data/soc-crosswalk.csv`](data/soc-crosswalk.csv) | **Every** BLS SOC 2018 occupation (867 codes) mapped to its nearest scored role | | [`data/changelog.md`](data/changelog.md) | Version history and notable movements | The crosswalk means any US occupation — even ones we don't score directly — resolves to a scored neighbor with a stated match quality. ## Use it in code **Python** ```python import requests data = requests.get("https://raw.githubusercontent.com/Jott2121/ai-career-threat-index/main/data/ai-career-threat-index.json").json() # Roles most exposed to autonomous agents specifically hot = sorted(data["roles"], key=lambda r: -r["subscores"]["agenticExposure"])[:10] for r in hot: print(f'{r["title"]:35} agentic={r["subscores"]["agenticExposure"]} overall={r["score"]}') ``` **JavaScript** ```javascript const url = "https://raw.githubusercontent.com/Jott2121/ai-career-threat-index/main/data/ai-career-threat-index.json"; const data = await fetch(url).then(r => r.json()); const rising = data.roles.filter(r => r.score - (r.historicalScores["Q2 2026"] ?? r.score) >= 3); ``` **R** ```r roles JSON + CSV + crosswalk + quarterly report python3 pipeline/generate_svgs.py # README charts pytest tests/ # 30+ integrity checks (CI runs these on every push) ``` Historical quarters (Q1 2025 – Q2 2026) were published under methodology v1 (50/30/20 weighted composite, no agentic factor) and are retained as published. Quarterly moves larger than ±8 points require a stated restatement reason — see the [quarterly report](reports/2026-q3-winners-losers.md). ## Cite it ``` The MeritForge Team (2026). AI Career Threat Index v2026.3. https://github.com/Jott2121/ai-career-threat-index (DOI: 10.5281/zenodo.21227026) ``` See [`CITATION.cff`](CITATION.cff) for BibTeX and more formats. When citing in editorial content, a link to this repo or the [interactive explorer](https://jott2121.github.io/ai-career-threat-index/) is appreciated. ## Contributing Disagree with a score? The sub-scores and rationales are published precisely so you can attack them: [open an issue](../../issues) naming the role, the factor, and the evidence. Better adoption data for an industry is especially welcome. Accepted corrections land in the next quarterly review with attribution. ## About Maintained by [Jeff Otterson](https://github.com/Jott2121) / The MeritForge Team. The interactive tool also lives at [meritforgeai.com](https://www.meritforgeai.com/tools/ai-career-threat-index/). --- ⭐ **If this dataset is useful, star the repo** — stars are how researchers and journalists find it. ## Source & license This open-source MCP server is cataloged on AgentStack and links to its original source — we do not rehost the code. - **Author:** [Jott2121](https://github.com/Jott2121) - **Source:** [Jott2121/ai-career-threat-index](https://github.com/Jott2121/ai-career-threat-index) - **License:** MIT - **Homepage:** https://jott2121.github.io/ai-career-threat-index/ Install and usage instructions live in the source repository linked above. ## Pricing - **Free** — Free ## Security capabilities Automated source analysis of v0.1.0 — what this tool can access: - **Network access:** yes - **Filesystem access:** no - **Shell / process execution:** no - **Environment & secrets:** no - **Dynamic code execution:** no *"Yes" means the capability is present in the source — more access means more to trust, not that it is unsafe.* ## Versions - **0.1.0** — security scan: passed — Imported from the upstream source. ## Links - Listing page: https://agentstack.voostack.com/l/mcp-jott2121-ai-career-threat-index - Seller: https://agentstack.voostack.com/s/jott2121 - Browse the marketplace: https://agentstack.voostack.com/browse --- Listed on AgentStack — the marketplace for AI agent skills and MCP servers. Every listing is security-reviewed. Creators keep 70%.