Developer Moves All Prompts to Skills: AI Coding Workflow v3 Explained
rseroter · x · 2026-10-26
Developer Helder Burato Berto published "My AI Coding Workflow, v3", tracing the evolution from a hand-written SPEC.md (v1) through TracerKit (v2) to v3, where every step that used to be a prompt is now a skill and every "please remember to" is a check that fails.
Key points:
- Why he dropped TracerKit: a separate tool meant its own release cycle; moving the spec flow into the agent-skills repo makes any change a single commit.
- v3 combines the best of two repos: from Matt Pocock's skills he took the flow shape — grill-me (interviews until the plan has no open branches), to-spec and to-tickets with tracer-bullet tickets and blocking links, a phased debugging loop, plus tdd/prototype/domain-modeling/retro; from Addy Osmani's agent-skills he took engineering depth per step.
- The pipeline runs Plan (triage→grill→to-spec→to-tickets), Build (build→review), Ship (to-pr→retro).
- Core ideas carried over: interviews before specs, tracer-bullet tickets, verification before a task counts as done.
- His summary: "The spec is still at the center... The prompts I kept retyping are skills, and the reminders I kept giving are checks."
More from coding & agent
- Security tip: use AI agents to build a NIST CSF knowledge base for incident response — prasanna_says · 2026-10-26
- Google Cloud Adds Self-Learning Context Enrichment to Knowledge Catalog for AI Agents — rseroter · 2026-10-26
- Cloudflare Previews Spend Limits for Workers and Durable Objects, Early Access Open — ritakozlov · 2026-10-26
- Shopify design lead on AI-era workflow: designers are becoming full-stack editors — KlausCodes · 2026-10-26
- Nace AI launches Drex for Search, cutting agent steps 62% and tokens 56% — testingcatalog · 2026-10-26
- Ultrametric open-sources Open Startup Repo to let AI agents run startup operations — garrytan · 2026-10-26