---
name: ask-jason
description: Access Jason's long-form writing on builders, agents, and hardware-AI loops. Use when the user wants to read, summarize, search, or stay updated on Jason's essays and thinking at jason.ltd. Fetches the latest essays and detects new ones via a static JSON feed — no API key required.
---

# Ask Jason

Jason is a builder-PM and co-founder of HeyGo (an AI snowboarding coach). He writes long-form essays on agents, hardware-AI loops, product sense after 12 years of shipping, and knowledge systems at https://jason.ltd/writing.

This skill lets you read his writing and stay current with new essays. No API key, no MCP server.

## Data sources

| What | Where | Format |
|------|-------|--------|
| Essay list + update detection | `https://jason.ltd/writing/feed.json` | JSON Feed 1.1 |
| Full essay (HTML) | `https://jason.ltd/writing/{id}` | HTML |
| Site + writing overview | `https://jason.ltd/llms.txt` | text |

## How to use

**Read the latest essay**
1. `GET https://jason.ltd/writing/feed.json`
2. `items` are newest-first — take `items[0]`.
3. Fetch its `url` for the full text.

**"What has Jason written about X?"**
1. GET the feed, scan each item's `title` / `summary` / `tags`.
2. Fetch matching essays' `url` for detail.

**Track / follow updates**
1. Remember the `id` (or `date_published`) of the newest item you've seen.
2. On a later check, GET the feed again. Any item with an unseen `id` or newer `date_published` is a new essay.
3. There is no push — poll the feed when the user asks "anything new from Jason?".

## Notes
- The feed is static and cache-friendly; polling is cheap.
- `summary` is Jason's own excerpt — use it for quick triage before fetching full text.
- Essays are mostly English; each item's `language` field marks the language.