Agent Workflows: Specs and Tickets vs Direct Prompting

X postAndrei MaximOpinion

Andrei Maxim argues that writing specifications and splitting work into tickets is outdated with modern models like Sol 5.6 and Opus 5.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.

Key points

  1. Andrei Maxim's post responds to the release of mattpocock/skills v1.3, which includes spec-and-ticket workflows driven by subagents.
  2. Maxim claims that with models such as Sol 5.6 and Opus 5, manually creating specs and tickets is no longer necessary.
  3. For larger initiatives, he advises collaborating with the agent to split the work into sequential, shippable PRs.
  4. He notes that context compaction issues have diminished, making long agent sessions far less prone to quality degradation.

Why it matters

As coding agents and underlying LLMs become more capable, the necessity of rigid workflow frameworks and meta-prompts is increasingly questioned in favour of simpler, direct interaction.

Analysis

The discussion highlights a divide in the software engineering community. While Maxim favours lean, conversational task delegation enabled by advanced model capabilities, other developers counter that specs and ticket architectures remain essential for team collaboration, codebase longevity, and large enterprise environments.

Original post

Andrei Maxim

@rockatanescu · 5 October 2026

I don't think writing a spec and splitting it into tickets is how you should use agents since Sol 5.6/Opus 5. Just tell the agent what you want to do and it will do it. If the task is large, discuss with the agent how to split it in several shippable PRs and do it one at a time. t.co/HPcUoLTcdi

164 likes38 replies123 bookmarks24.7k views

View on X
More posts in the thread (1)
  1. Compaction used to be really bad, but today you don't need to worry about it, especially if you don't rely on some specific content that only appears in the middle of the of the previous context window. This thread is from two days ago. I never felt I got into the "dumb zone". t.co/3Pb7VXO8Ow
Selected replies (5)
  • @doodirock @rockatanescu You’re not using these methods for the first time date implementation. You’re doing it for the longevity of your application and your teammates ability to work on it with you.
  • @scottw @rockatanescu @sqs This has been my experience as well. I have retired all my workflow skills.
  • @Skunkmonkey1243 @rockatanescu Vibe coding works for small projects. But when your codebase is tens of millions of lines of code, you're asking for trouble. You need some form of map, be that SDD, a lookup table, an SQLite DB, or whatever you choose. Vibe coding is not an option for enterprise.
  • @th1agofm @rockatanescu He's just milking what he can. Just prompt it, agree!
  • @jlydev @rockatanescu AI skills ~= astrology for developers

Mentioned links

Summary generated by AI (Gemini 3.8 Flash, 6 October 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

  • X post:MIKE

    Prompting Claude Opus 5: New Rules for Developer Workflows

    MIKE highlights a breakdown by Rahul (@sairahul1) covering Anthropic's official prompting guide for Claude Opus 5, released on 24 July 2026. The guide details how Opus 5 requires a fundamental shift in prompting techniques compared to Opus 4.8.

    1.8kbookmarks159.6kviews

    AI & LLMs· Resource

  • X post:nicco

    Beyond Prompting: Agentic Workflows as Execution Graphs

    Citing a Stanford lecture by Andrew Ng, author nicco highlights why zero-shot prompt engineering is reaching its limits and being replaced by autonomous execution graphs, supported by a guide on moving away from linear agent chains.

    275bookmarks25.1kviews

    AI & LLMs· Opinion

  • X post:Magnus Müller

    Agency: AI Agent Proposes Tasks Instead of Waiting for Prompts

    Magnus Müller has introduced Agency, an approach that reverses traditional prompting: instead of waiting for user instructions, the agent proactively identifies pending work for user approval, using a me.md file as a personal context base.

    2.3kbookmarks227.7kviews

    AI & LLMs· Tool

  • X post:Mahax

    Moving Beyond Prompts to Loops and Graphs in AI Systems

    Citing Andrew Ng, Mahax highlights the shift from single-shot prompting to agentic loops and graphs. An accompanying guide outlines the architecture of AI agents, self-evaluating feedback loops, and graph-based execution pipelines using Claude.

    11.8kbookmarks1mviews

    AI & LLMs· Resource

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.