# QuantGist AI Visibility Benchmark

Last updated: 2026-05-18

This benchmark evaluates how ready the QuantGist portal is for reliable AI discovery, ingestion, citation, and reuse by third-party LLMs and AI agents.

## Score

Current score: **76 / 100**

| Area | Weight | Current | Notes |
|---|---:|---:|---|
| LLM discovery | 20 | 15 | `llms.txt` exists on the portal and API docs exist, but asset-level discovery should be linked from every relevant surface. |
| Machine-readable API surface | 20 | 17 | FastAPI OpenAPI and API LLM references exist. Add per-endpoint Markdown mirrors and examples. |
| Markdown content portability | 15 | 10 | Blog content is MDX and backend docs are Markdown. Portal pages need clean Markdown exports. |
| AI asset catalog | 15 | 13 | Catalog now includes the existing MCP server and applicable distribution assets; it should be kept in sync with real product/admin data. |
| Metadata and structured data | 10 | 7 | Site and article JSON-LD exist. Add Product, SoftwareApplication, APIReference, FAQ, and Dataset schemas where relevant. |
| Accessibility and alt text | 10 | 6 | Basic metadata exists. Need an image/alt-text audit and descriptive media policy. |
| Evaluation and monitoring | 10 | 8 | This benchmark gives a baseline. Next step is an automated CI check and monthly crawl report. |

## Current Strengths

- Portal-level `llms.txt` exists at `https://quantgist.com/llms.txt`.
- API-specific LLM references exist at `https://api.quantgist.com/llms.txt` and `https://api.quantgist.com/llms-full.txt`.
- FastAPI exposes OpenAPI at `https://api.quantgist.com/openapi.json`.
- Backend docs are Markdown-based.
- Blog/insights content is MDX-based with article metadata and JSON-LD.
- Sitemap and robots routes exist.
- Product already has AI-relevant assets: sentiment scoring, impact scoring, symbol tagging, webhook delivery, SDKs, an MCP server, CLI, bots, widgets, MT5 bridge, and trading integrations.

## Main Gaps

- Portal pages such as `/platform`, `/solutions`, `/features`, `/pricing`, and `/docs` do not yet expose clean `.md` equivalents.
- `llms.txt` is not yet generated from source-of-truth content, so it can drift.
- AI and distribution assets are not yet shown as a product catalog inside the user-facing portal.
- MCP server exists in the ecosystem workspace, but package publishing, public docs, and production setup should be confirmed.
- Several distribution assets are complete or scaffolded but still need publishing, deployment, or marketplace submission status confirmed.
- OpenAPI, Markdown docs, SDK README files, and web docs may drift because there is no single docs build pipeline.
- Static images and screenshots need explicit alt-text and text fallbacks where they contain product or schema information.
- There is no automated AI-visibility score in CI.

## Roadmap

### Phase 1: Discovery Baseline

Target: 1-2 days

- Keep `/llms.txt`, `/ai-assets.json`, `/ai-assets.md`, and `/ai-visibility-benchmark.md` public.
- Add links to AI assets from the footer, docs page, and developer docs.
- Include `llms.txt` in sitemap or docs navigation.
- Confirm `robots.txt` allows all public AI discovery files.

Expected score after phase: **78 / 100**

### Phase 2: Markdown Mirrors

Target: 2-4 days

- Add public Markdown versions for core portal pages:
  - `/platform.md`
  - `/solutions.md`
  - `/features.md`
  - `/pricing.md`
  - `/docs.md`
  - `/faq.md`
- Add Markdown mirrors for each insight article or expose original MDX safely.
- Add canonical links between HTML and Markdown versions.

Expected score after phase: **85 / 100**

### Phase 3: Structured AI Metadata

Target: 2-3 days

- Add JSON-LD for `Product`, `SoftwareApplication`, `APIReference`, `Dataset`, `FAQPage`, and `BreadcrumbList`.
- Add field-level metadata for core schemas: event, calendar event, webhook delivery, watchlist, usage.
- Add a public schema glossary in Markdown and JSON.

Expected score after phase: **90 / 100**

### Phase 4: AI Assets as Products

Target: 3-5 days

- Build a public `/ai-assets` page using the same source data as `ai-assets.json`.
- Add individual pages for:
  - Market Intelligence API
  - Sentiment Analyzer
  - Impact Scorer
  - Symbol Tagger
  - Webhook Delivery Agent
  - Python SDK
  - JavaScript SDK
  - QuantGist MCP Server
- Add distribution asset pages for:
  - qgist CLI
  - QuantGist MT5 Bridge
  - Micro-Apps and Widgets
  - Telegram and Discord Alert Bots
  - freqtrade Protection Plugin
  - TradingView Webhook Relay
- Add lifecycle status, access requirements, examples, pricing tie-ins, and integration snippets.

Expected score after phase: **94 / 100**

### Phase 5: Agent-Native Tooling

Target: 1-2 weeks

- Promote the existing QuantGist MCP server as an AI asset:
  - `get_upcoming_events`
  - `get_events_range`
  - `check_safe_to_trade`
  - `get_economic_calendar`
  - `get_event_detail`
- Add write tools later with stricter auth:
  - `create_watchlist`
  - `register_webhook`
- Add package publishing status, tool manifests, example prompts, and agent integration guides.

Expected score after phase: **97 / 100**

## Measurement Plan

Run this benchmark monthly and after major docs releases.

Minimum checks:

- Does `https://quantgist.com/llms.txt` return HTTP 200 and `text/plain`?
- Does it link to API docs, OpenAPI, full LLM reference, asset catalog, pricing, and signup?
- Does `https://quantgist.com/ai-assets.json` return valid JSON?
- Does every asset have `id`, `type`, `status`, `name`, `description`, and `docs_url` or equivalent?
- Does OpenAPI return HTTP 200 and include every public endpoint?
- Do core docs pages have Markdown alternatives?
- Do pages include canonical URLs and useful titles/descriptions?
- Do article pages include Article JSON-LD?
- Do product/docs pages include relevant schema.org JSON-LD?
- Do images with meaningful content have alt text or text equivalents?

## Target

QuantGist can realistically reach **90+ / 100** with Markdown mirrors, richer structured metadata, and an automated benchmark. It can reach **95+ / 100** once AI assets are exposed as product pages and the existing MCP server has public launch docs, package publishing, and integration examples.