Agentic Development
Building with agents. Building tools for agents. A field journal from the practice.
01. Writing
The Series
Field notes from a daily practice with AI coding agents. Patterns, tools, sharp edges. Evidence-based, no fluff.
Shannon v1.2.0: Multi-Stage Agentic Work as a Sequence of Provable Steps
A Claude Code plugin that refuses to say 'done' until the evidence is on disk. 36 skills, 11 agents, 22 commands, 7 hooks — and a doctor that reads its own contract.
Sublimate: Three Tiers for Distilling Workflows From Your Own Sessions
Anthropic shipped Dynamic Workflows in Claude Code this week. The harder question is which of your hand-rolled patterns belongs as a Workflow, which belongs as a Skill, and which belongs as a Subagent.
When Your Drafter Doubles the Body: Building the WithAgents Content Plugin
The clear step in our LinkedIn drafter never actually cleared. The new body pasted on top of the old one, the editor doubled, and the bug only showed up on the 34th run. Here is what we shipped instead.
Orbit: Find Drift Between What You Asked and What Actually Shipped
Plans are evidence, not history. Mining the JSONL session transcripts you already have on disk to compare intent against claim against codebase truth.
02. Products
What we ship
ValidationForge proves that code works. Anneal proves the plan works before any code gets written. Two flagships, both production-grade. The other cards are companion tooling from the series.
Get the next field note before it ships.
4,206 engineers, operators, and weirdos. No spam, ever.