14 posts on tutorials for AI agents that send and receive email.
6 August 2026 · 3 min read
How to write the golden thread harness in code: fixtures, assertions on actions rather than wording, a fake mail client, and the adversarial cases worth pinning.
6 August 2026 · 4 min read
A wrong reply has five common causes and each leaves a different trace. How to find which one it was, in order, without reading message bodies out of a log.
4 August 2026 · 3 min read
Define email as typed tools, keep the loop bounded, and put the untrusted message where it cannot reach a privileged call. A working shape with the parts that matter explained.
4 August 2026 · 3 min read
The whole surface is an HTTP API with a bearer token. What the four calls look like in Go, Ruby, and PHP, and the three things to get right whatever you write it in.
3 August 2026 · 4 min read
Some clients cannot store an API key, so they authorize as you instead. How the OAuth flow works, what the scopes cover, why a token needs an agent header, and when to prefer a key.
3 August 2026 · 4 min read
Not everything runs on the edge. The same agent as a Node service: webhook endpoint, a queue, per-thread locking in Postgres, and the deployment details that decide whether it survives a burst.
1 August 2026 · 4 min read
Wire an inbox to an automation tool: receive with a webhook, decide with an LLM step, reply through one HTTP call, and the four things to get right so it does not double-send.
1 August 2026 · 4 min read
Wire an inbox into a LangChain or LangGraph agent: MCP where the adapter exists, plain tools where it does not, and where the email work belongs in a graph.
30 July 2026 · 3 min read
A working TypeScript agent with its own inbox: install, send, read the thread, wait for a reply, handle refusals, and receive inbound mail without a polling loop.
30 July 2026 · 4 min read
Connect a GPT-based agent to a real inbox: MCP where the tooling supports it, plain function tools where it does not, and the guardrails to set before it emails anyone.
30 July 2026 · 4 min read
The CLI is the fastest way to see what an agent is doing: list its mail, read a thread, send a test message, and script the checks worth running in CI.
27 July 2026 · 4 min read
A working Python agent with its own inbox: install the SDK, send a first message, read threads, wait for replies, handle refusals, and receive inbound mail without a polling loop.
27 July 2026 · 4 min read
A working email agent as a Durable Object: connect the MCP server, react to inbound mail through a webhook, hold a thread across days, and keep the whole thing idempotent.
25 July 2026 · 5 min read
Connect Claude to a real inbox in a few minutes with an MCP email server: config, the first send, holding a conversation, and the limits worth knowing before you let a model email people.