Tuning the evals that grade this site's agentsShowing how to connect an MCP client to this site at /connectShipping armature, a Claude Code plugin that works an epic end to end
Public MCP tool calls
Chat sessions
Requests from agents
AI Gateway requests
Edge compute requests
Latest · Agentic engineering
Agents that repair the plan as they work
Armature v0.9.0 lets an agent fix the tickets it works from: broken text, a missing blocker, a dependency nobody wrote down. Why a plan needs that, how dependency links and the prose that states them stay in step, and the limits that make it safe to allow without approval.
· 6 min read
About
Ryan Lindsey, Senior Engineering Manager, Agentic Engineering.
Agentic engineering is making engineers dramatically faster. I build the instruments that let the organization around them keep pace. Nineteen years in software, a decade of it leading engineering teams: I own incident management for an entire engineering organization, I build the agentic tooling that other managers choose to adopt, and I still ship the code. The habit underneath all of it is to measure a process before improving it, because the constraint is not always where it is assumed to be.
More writing
All posts →· Agentic engineering
Chaining skills across plugins
Skills from two Claude Code plugins, armature and superpowers, now run as one pipeline from a design conversation to a stack of reviewed pull requests, and neither plugin edits the other. How the handoffs work, what they cost, and six rules for writing skills that compose.
14 min →
· Building in the open
Choosing a Workflow over a Queue
A long, metered inference call on this site was deferred off its request and canceled at thirty seconds on every real run, and no test could see it. Cloudflare documents a Queue as the remedy, and a Workflow was chosen instead. This is why, told through two Workflows that now run here, with the decision rule I would hand another engineer and the three defaults I would override on the first day.
17 min →
· Building in the open
Verifying a change no test can reach
The search on this site could not find a phrase that appears in ordinary prose on five of its own pages. Reranking fixed that in measurement, was switched on, and the page still returned nothing. What each verification step confirmed, what a person searching actually got back, and the checks that close the gap.
7 min →