Skip to content
View antonyevans's full-sized avatar

Block or report antonyevans

Block user

Prevent this user from interacting with your repositories and sending you notifications. Learn more about blocking users.

You must be logged in to block users.

Maximum 250 characters. Please don’t include any personal information such as legal names or email addresses. Markdown is supported. This note will only be visible to you.
Report abuse

Contact GitHub support about this user’s behavior. Learn more about reporting abuse.

Report abuse
antonyevans/README.md

Antony Evans

Cambridge mathematician. YC-backed founder. Currently building at the agentic commerce layer — where AI agents shop, transact, and operate autonomously.

Background: maths → strategy consulting (Oliver Wyman, Bain) → startups → COO. I've raised $42M in venture capital, scaled a company to $50M revenue, and shipped software used by 500k+ people. Now building the infrastructure layer for agent-native commerce.


🔨 Active Projects

A shopping CLI built for AI agents, not humans.

Humans configure it once — budgets, approval thresholds, spending policy. Agents use it to shop autonomously. The core primitive is a mandate: a signed, revocable spending delegation that encodes purchase policy. Agents reference a mandate ID on every order; they never see payment credentials.

  • pip install shop-cli — on PyPI, Python 3.9+
  • Every command returns JSON. All mutations take --idempotency-key. Semantic exit codes — agents branch on 3 (mandate violation), 5 (low confidence), 6 (retry-safe network error).
  • Stack: Python 3.12, typer, pydantic v2, httpx (async), Ed25519 mandate signing, SQLite

Production-ready Claude Code skills, on demand.

Submit a spec — get back a complete skill with SKILL.md, reference docs, evals, and scripts, built to the same standard as skills I run in production. The underlying skill architecture is open-source: a growing library of domain-specific skills covering marketing, product, legal, operations, and more.

Multi-agent pipeline: Claude Code + MCP → recipe generation → cart population → phone notification. No human in the loop after the weekly trigger.

Android symptom checker — 500k+ downloads, medical content partnership with Harvard Medical School. Built in Java; still the highest-distribution thing I've shipped.


⚙️ Technical Stack

AI / Agents: Python, Claude SDK (Anthropic), Claude Code, MCP (Model Context Protocol), multi-agent orchestration, LangChain, RAG pipelines, prompt engineering

Data: SQL / BigQuery, Tableau, financial modeling (Excel/Python)

Infrastructure: Git, Jupyter, VS Code, REST APIs, Stripe


📍 Currently Building

  • shop-cli — agent-native shopping protocol, pip install shop-cli (github.com/antonyevans/shop-cli)
  • Finch Toys — DTC dropship store with agentic commerce infrastructure underneath
  • Agent Engineer Master — skill-as-a-service platform for Claude Code (agentengineermaster.com)

📫 Connect

Pinned Loading

  1. anthropic-cookbook anthropic-cookbook Public

    Forked from anthropics/claude-cookbooks

    A collection of notebooks/recipes showcasing some fun and effective ways of using Claude.

    Jupyter Notebook

  2. browser-use browser-use Public

    Forked from browser-use/browser-use

    Make websites accessible for AI agents

    Python 1

  3. grocery-list grocery-list Public

    Python

  4. highfidelityassets highfidelityassets Public

    Place to store avatar attachments for high fidelity

    JavaScript