ARGUS

Evidence-led tracking for emerging AI technologies.
View the live tracker · Read the FAQ

ARGUS is an evidence-led technology hype and maturity tracker. It compares public attention with observable adoption and maturity signals, then places a technology on an explainable five-stage lifecycle.

The included showcase follows a growing portfolio of AI technologies, including agent harnesses, browser agents, LangGraph, and Model Context Protocol (MCP). It runs on a dependency-free Python service, SQLite, public GitHub/Hacker News/Google News metadata, and a static web interface.

What is included

• Public overview, technology detail pages, methodology, and FAQ
• Admin console for source health, evidence review, analysis runs, and technology lifecycle management
• 52-week historical baseline from public metadata
• OpenRouter-assisted candidate discovery and editable draft-profile preparation
• Docker Compose, GitHub Container Registry, and VM deployment support

Local run

python3 scripts/refresh_data.py
python3 scripts/run_local.py --port 8000

The public interface is available at http://127.0.0.1:8000; /admin is the operations console and /faq explains the model.

Docker Compose

cp .env.example .env
docker compose up --build

ARGUS_ADMIN_TOKEN is required for shared environments. Compose stores operational state in the named argus_data volume. docker compose down -v removes that state.

Configuration

Variable	Purpose
ARGUS_ADMIN_TOKEN	Admin API token. Defaults to argus-local only outside Compose.
GITHUB_TOKEN	Optional fine-grained read-only GitHub token for more reliable collection.
OPENROUTER_API_KEY	Optional key for weekly discovery and admin draft-profile preparation.
ARGUS_CLOUDFLARE_ANALYTICS_TOKEN	Optional Cloudflare Web Analytics site token; public pages only.
ARGUS_LLM_MODEL	openai/gpt-5.6-luna by default; the single OpenRouter model for both discovery and draft-profile preparation.

Data boundaries

ARGUS retains public metadata, feed titles, short permitted summaries, and source links—not article bodies. Historical signals use GitHub activity and dated Hacker News discussion; current coverage also includes Google News RSS. Repository activity is a limited adoption proxy, not production evidence. The result is a research estimate, not investment advice or an official Gartner classification.

The committed showcase cache was generated on 2026-07-27. scripts/refresh_data.py refreshes it from the configured public sources.

Validation

PYTHONPATH=backend/src python3 -m unittest discover -s backend/tests -v

Documentation

• Methodology
• HTTP API
• Operations runbook
• Technology discovery
• Source, confidence, and lifecycle resilience plan
• Production deployment
• Runtime technology profiles
• Source and maintenance scripts
• Architecture plan

Repository layout

ARGUS/
├── backend/     API, collection, inference, storage, and tests
├── config/      versioned technology profiles and source policy
├── data/        local state policy and committed fixtures
├── docs/        methodology, operations, API, and deployment reference
├── frontend/    public and admin web interfaces
├── infra/       container and reverse-proxy assets
└── scripts/     local collection and maintenance commands

ARGUS is independent of Gartner. Public copy uses “technology hype and maturity” terminology and does not reproduce proprietary Gartner graphics.