Claude Code: Using a Dedicated Subagent for Visual Project Maps During Overnight Runs

X postVoxTutorial

Vox (@Voxyz_ai) outlines a prompt workflow for Opus 5.5 in Claude Code that configures a dedicated subagent called 'project-map'. The agent runs in the background and maintains an interactive HTML dashboard showing progress and blockers during long autonomous coding sessions.

Key points

  1. The 'project-map' subagent is configured in ~/.claude/agents using Opus at medium effort with user memory enabled.
  2. Styling preferences (dark/light theme, accent colour) are collected on first run and stored across projects.
  3. The map outputs as a standalone HTML file in a git-ignored '.project-map/' folder, openable with a double-click.
  4. The dashboard splits the codebase into areas and marks statuses: done, in progress, not started, or stuck (including what is blocking it).
  5. A global rule in ~/.claude/CLAUDE.md instructs the main session to trigger map updates before unsupervised stretches and after milestones.
  6. Community replies note that tracking git history alone can be misleading without linking statuses directly to CI or test suite outputs.

Why it matters

Allowing AI models like Opus 5.5 to run unattended overnight often leaves developers with extensive commit logs to decipher. Delegating visual project tracking to a specialised subagent separates reporting from development, giving developers an instant visual overview upon returning to the project.

Analysis

The technique takes advantage of subagent orchestration in Claude Code. By offloading UI-based status tracking to a background task, the main coding agent remains focused on high-effort development. However, as peers noted in discussion, deriving progress primarily from git commits risks reporting tasks as 'done' even if runtime tests fail; integrating automated test reports directly into the map cards addresses this vulnerability.

Original post

Vox

@Voxyz_ai · 6 October 2026

Every time you let Opus 5.5 work on a project overnight, have it vibe-code a quick 𝗽𝗿𝗼𝗷𝗲𝗰𝘁 𝗺𝗮𝗽 like this first. In the morning, you'll see at a glance what's done and what's left. Then give it 𝗮 𝘀𝘂𝗯𝗮𝗴𝗲𝗻𝘁 𝘁𝗵𝗮𝘁 𝗼𝗻𝗹𝘆 𝗱𝗿𝗮𝘄𝘀 𝗽𝗿𝗼𝗷𝗲𝗰𝘁 𝗺𝗮𝗽𝘀: → t.co/8k56bDtiE2

1.5k likes53 replies3.6k bookmarks115.6k views

View on X
Selected replies (5)
  • @ihurricanez @Voxyz_ai A subagent whose only job is drawing maps. Finally an intern with one clear task lol
  • @krl_grn @Voxyz_ai Love this. I ended up going one step further: instead of a map drawn on top, the agents work off a task board they run themselves. Every task has a live status, blocked ones say what they're waiting on, and in the morning I just ask "what got done?" in any chat. t.co/cQanlYWCbS
  • @burnnnnnnt @Voxyz_ai @SaveToList
  • @ItsGoharr @Voxyz_ai overnight coding with a babysitter agent watching
  • @scrblanc @Voxyz_ai the map reads git history, so 'done' tracks commit messages, not whether the code runs. bind each part's status to CI or test results, or you've built a dashboard that's confidently wrong during the exact overnight stretch it exists to cover.

Mentioned links

Summary generated by AI (Gemini 3.8 Flash, 7 October 2026). It may contain errors – the original source is authoritative.

Related entries

Similar in content, found by the AI search.

  • X post:Mr. Buzzoni

    Claude Loop Engineering: Structured Agent Loops for Autonomous Overnight Runs

    Mr. Buzzoni (@polydao) outlines a file-based harness and directory architecture for Claude Code that enables AI agents to execute overnight tasks autonomously. Using clear contracts, verification rubrics, state tracking, and documented receipts, the setup is designed to run predictably without manual night-time interventions.

    373bookmarks29.7kviews

    AI & LLMs· Resource

  • 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:Eric Buess

    Creative Visualisations and 3D Projects Powered by Claude Opus 5.5

    Eric Buess curates a collection of community demonstrations showcasing how Claude Opus 5.5 is used for programmatic animations, 3D modelling, and video direction, frequently generating code rather than using image models.

    337bookmarks79.9kviews

    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.