Matt Pocock Tests Multi-Agent Workflow via /implement-spec

X postMatt PocockOpinion

Matt Pocock is experimenting with a skill called /implement-spec that executes specifications and tickets using concurrent subagents and automated review.

Key points

  1. The /implement-spec skill accepts a specification document and a set of tickets as input.
  2. A dedicated subagent investigates the existing codebase before implementation begins.
  3. Tickets are implemented concurrently across multiple subagents to achieve maximum parallelism.
  4. The final codebase is reviewed against the original specification before cleanup.
  5. Pocock places this step in the pipeline /wayfinder -> /to-spec -> /to-tickets -> /implement-spec, noting that higher concurrency increases token usage.

Why it matters

Multi-agent coding workflows shift software engineering tasks from single sequential prompts to distributed, specialized execution pipelines. Pocock's experiment illustrates how separating research, implementation, and verification across subagents works in practice, alongside the trade-off of higher token consumption.

Analysis

By breaking the engineering workflow into discrete stages (/wayfinder to /to-spec, /to-tickets, and finally /implement-spec), Pocock treats agentic coding as a structured pipeline. Delegating codebase analysis and concurrent ticket execution to subagents speeds up delivery, but as Pocock points out, aggressive parallelization demands considerably more tokens, highlighting the persistent balance between speed and execution cost.

Original post

Matt Pocock

@mattpocockuk · 21 August 2026

I'm trying out an /implement-spec skill Essentially a multi-agent implementer that: - Takes in a spec and tickets - Does codebase research in a subagent - Implements all the tickets in subagents with maximum concurrency - Reviews the final code against the spec - Cleans up all

3.2k likes250 replies3.4k bookmarks471.9k views

View on X
More posts in the thread (3)
  1. @doodirock /wayfinder -> /to-spec -> /to-tickets -> /implement-spec
  2. @Ivanttyy More parallelization, but more tokens. So maybe not
  3. @srathsara1 Model, harness, effort?
Selected replies (5)
  • @srathsara1 @mattpocockuk He @mattpocockuk, I've been using this /implement-spec skill from the day you released it. just wanted to inform you on a issue I faced and I changed it a bit to fix it. Every time a sub-agent finish a development, the orchestrator runs tests and fill up the context window.
  • @doodirock @mattpocockuk Where does this fit in the current workflow? You would run after alignment via wayfinder instead of running /to-spec ?
  • @Ivanttyy @mattpocockuk is i better then simple ralph going through the tickets?
  • @bil0090 @mattpocockuk Thats going to be a GREAT one Most importantly, no em dashes? 👀
  • @SimplexForce @mattpocockuk 我有类似的实践，大概持续了两个星期。agent 会为准备持续投入的事情搭建一个团队。然后很出乎意料地这个团队模式对于长程运行发挥了极大作用。比如有个 subagent 专门做开发工作，每完成一个阶段性任务需要向主 agent 汇报工作，里面还包含使用了哪些 skill。主 agent 则在它的工作区维护一些工作报告

Mentioned links

Summary generated by AI (Gemini 3.8 Flash, 27 September 2026). It may contain errors – the original source is authoritative.

Related entries

Similar in content, found by the AI search.

  • X post:Matt Pocock

    Matt Pocock on AI Agent Workflows and Custom Skills

    Responding to common pitfalls in agent-driven software development, Matt Pocock outlines a structured workflow utilizing specialized commands such as /wayfinder, /to-spec, and /implement-spec.

    156bookmarks14kviews

    AI & LLMs· Opinion

  • Repository:mattpocock/skills

    mattpocock/skills: Modular Agent Skills for Real Software Engineering

    Matt Pocock shares a curated collection of agent skills taken directly from his own development setup. The project positions itself against rigid frameworks like GSD, BMAD, and Spec-Kit, as well as uncontrolled 'vibe coding'. Instead, it provides small, composable, and adaptable skills compatible with Claude Code, Codex, and other coding agents.

    247.2kstarsShell

    AI & LLMs· Tool

  • X post:Andrei Maxim

    Agent Workflows: Specs and Tickets vs Direct Prompting

    Andrei Maxim argues that writing specifications and splitting work into tickets is outdated with modern models like Sol 5.6 and Opus 5. Instead, developers should instruct the agent directly and break large tasks into shippable pull requests through dialogue.

    123bookmarks24.7kviews

    AI & LLMs· Opinion

  • Video

    Video:Matt Pocock

    Wayfinder: Planning Massive Software Projects Across Multi-Session AI Agents

    Matt Pocock introduces Wayfinder, an AI planning skill designed for coding agents to coordinate large-scale engineering work across multiple sessions. Rather than trying to cram broad architecture decisions into a single context window, Wayfinder maps complex goals into discrete decision tickets within standard issue trackers.

    AI & LLMs· Tutorial

Back to the knowledge base

Let's talk about your project

Get in touch

Locations

  • Mattersburg
    Johann Nepomuk Bergerstraße 7/2/14
    7210 Mattersburg, Austria
  • Vienna
    Ungargasse 64-66/3/404
    1030 Wien, Austria

Parts of this content were created with the assistance of AI.