Problem after update

[glebov]

Hello.
I used version of workflow, they after sub agents release a have updated it using command with @latest. Update finished successfully, but when i use claude code a see:

/spec-orchestrate is running… architecture refactoring

⏺ I'll start the orchestration of the architecture refactoring specification. Let me first load the context and analyze the current state.

⏺ Bash(npx @pimzino/claude-code-spec-workflow@latest get-tasks "architecture refactoring" --mode next-pending)
⎿  Error: tasks.md not found at .../.claude/specs/architecture refactoring/tasks.md

⏺ Let me check if the spec directory exists and what files are available:

⏺ filesystem-server - list_directory (MCP)(path: "....claude/specs")
⎿  [DIR] analytics-reorganization
[DIR] architecture-refactoring
[DIR] multi-level-thematic-analysis

⏺ I see the directory is named "architecture-refactoring" (with a hyphen). Let me try again:

⏺ Bash(npx @pimzino/claude-code-spec-workflow@latest get-tasks "architecture-refactoring" --mode next-pending)
⎿  No tasks found

⏺ Bash(npx @pimzino/claude-code-spec-workflow@latest get-tasks "architecture-refactoring" --mode all)
⎿  No tasks found

⏺ Let me check what files are in the architecture-refactoring directory:

⏺ filesystem-server - list_directory (MCP)(path: "...")
⎿  [FILE] commands.md
[FILE] design.md
[FILE] requirements.md
… +1 line (ctrl+r to expand)

⏺ The tasks.md file exists. Let me check if there are issues with the task parsing by reading the file directly:

⏺ filesystem-server - read_file (MCP)(path: "...tasks.md")
⎿  # Implementation Plan - Architecture Refactoring & Technical Debt Cleanup

 ## Task Overview
 … +371 lines (ctrl+r to expand)

⏺ Now let me check if agents are available for task execution:

⏺ Bash(npx @pimzino/claude-code-spec-workflow@latest using-agents)
⎿  false

⏺ Based on the tasks.md analysis, let me show the current status:

I try manuall enable sub agents, udpate spec flow, but error still exist.

How can i fix it?

[Pimzino]

What OS are you using?

…

On Tue, Aug 5, 2025 at 11:45 AM glebov ***@***.***> wrote: Hello. I used version of workflow, they after sub agents release a have updated it using command with @latest <https://github.com/latest>. Update finished successfully, but when i use claude code a see: /spec-orchestrate is running… architecture refactoring ⏺ I'll start the orchestration of the architecture refactoring specification. Let me first load the context and analyze the current state. ⏺ Bash(npx @***@***.*** get-tasks "architecture refactoring" --mode next-pending) ⎿ Error: tasks.md not found at .../.claude/specs/architecture refactoring/tasks.md ⏺ Let me check if the spec directory exists and what files are available: ⏺ filesystem-server - list_directory (MCP)(path: "....claude/specs") ⎿ [DIR] analytics-reorganization [DIR] architecture-refactoring [DIR] multi-level-thematic-analysis ⏺ I see the directory is named "architecture-refactoring" (with a hyphen). Let me try again: ⏺ Bash(npx @***@***.*** get-tasks "architecture-refactoring" --mode next-pending) ⎿ No tasks found ⏺ Bash(npx @***@***.*** get-tasks "architecture-refactoring" --mode all) ⎿ No tasks found ⏺ Let me check what files are in the architecture-refactoring directory: ⏺ filesystem-server - list_directory (MCP)(path: "...") ⎿ [FILE] commands.md [FILE] design.md [FILE] requirements.md … +1 line (ctrl+r to expand) ⏺ The tasks.md file exists. Let me check if there are issues with the task parsing by reading the file directly: ⏺ filesystem-server - read_file (MCP)(path: "...tasks.md") ⎿ # Implementation Plan - Architecture Refactoring & Technical Debt Cleanup ## Task Overview … +371 lines (ctrl+r to expand) ⏺ Now let me check if agents are available for task execution: ⏺ Bash(npx @***@***.*** using-agents) ⎿ false ⏺ Based on the tasks.md analysis, let me show the current status: I try manuall enable sub agents, udpate spec flow, but error still exist. How can i fix it? — Reply to this email directly, view it on GitHub <#37>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/BBYICLYB7JJ46SOU5KAGQ5L3MCDL7AVCNFSM6AAAAACDESJSJ6VHI2DSMVQWIX3LMV43ERDJONRXK43TNFXW4OZYGY4DONRVGY> . You are receiving this because you are subscribed to this thread.Message ID: ***@***.***>

[glebov]

Mac OS

[glebov]

And there are no directory with sub agents in my folder - but during installation terminal wrote, that they are enabled and ready to use.

[syafiqshamsuddinTSB]

Me also .. i think in the latest version something broke, but if i use git bash as terminal (i’m using windows) then those agents will only populate.

[syafiqshamsuddinTSB]

Also i had to edit the spec config json to make using agent: true

[Pimzino]

I have released an update today to resimplify the workflow as I overcomplicated it too much. Try the new update that has no dependencies on any spec-config.json etc.

[glebov]

Thank you.
I have installed globally ( npm i -g @pimzino/claude-code-spec-workflow), but should i delete previous installation, which i did in project directory?

[Pimzino]

Yes I would recommend so! Although technically it should back up that folder and create a fresh one

…

On Wed, 6 Aug 2025 at 16:40, glebov ***@***.***> wrote: Thank you. I have installed globally ( npm i -g @pimzino/claude-code-spec-workflow), but should i delete previous installation, which i did in project directory? — Reply to this email directly, view it on GitHub <#37 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/BBYICL6N3LPF5JYUKIKAIT33MIOWVAVCNFSM6AAAAACDESJSJ6VHI2DSMVQWIX3LMV43URDJONRXK43TNFXW4Q3PNVWWK3TUHMYTIMBSGI2TCMQ> . You are receiving this because you modified the open/close state.Message ID: <Pimzino/claude-code-spec-workflow/repo-discussions/37/comments/14022512@ github.com>

[glebov]

Why i see this problem? And how to fix?

/spec-orchestrate is running… architecture refactoring task 4.2

⏺ I'll orchestrate the execution starting from task 4.2 of the architecture refactoring specification. Let me first load the current state and context.

⏺ Bash(npx @pimzino/claude-code-spec-workflow@latest get-tasks architecture-refactoring --mode next-pending)
⎿  No tasks found

⏺ Bash(npx @pimzino/claude-code-spec-workflow@latest get-tasks architecture-refactoring --mode all)
⎿  No tasks found

⏺ Let me check if the specification exists and load the tasks file directly:

⏺ filesystem-server - read_file (MCP)(path: "/Users/ivan/Dev/rag_text_classifier/.claude/specs/architecture-refactoring/tasks.md")
⎿  # Implementation Plan - Architecture Refactoring & Technical Debt Cleanup

 ## Task Overview
 … +371 lines (ctrl+r to expand)

⏺ Now let me check if agents are available and load the context documents:

⏺ Bash(npx @pimzino/claude-code-spec-workflow@latest using-agents)
⎿  Error: - Analyzing project...
✔ Project analyzed: /Users/ivan/Dev/rag_text_classifier
- Starting fresh installation update...
Failed to delete .claude directory: [Error: EACCES: permission denied, unlink '/Users/ivan/Dev/rag_text_classifier/.claude/commands/multi-level-thematic-analysis/task-17.md']
{
errno: -13,
code: 'EACCES',
syscall: 'unlink',
path: '/Users/ivan/Dev/rag_text_classifier/.claude/commands/multi-level-thematic-analysis/task-17.md'
}
Fresh installation update failed: Error: Failed to delete existing .claude directory. Update cancelled for safety.
… +24 lines (ctrl+r to see all)

⏺ Let me load the requirements and design documents directly to get the context:

[glebov]

Agents enabled, folder agents exists, but in logs:
Since agents aren't available, I'll implement task 4.2 directly. Let me first examine the current A/B testing code in app.py to understand what needs to be extracted:

How can i fix it? I want to use sub agents ))

[glebov]

And... i have detected that now i have two backup folders and .claude is clear.
Maybe it is a bug, because it works without my approval and i dont understand why and the reason (

[Pimzino]

Brother, I would recommend start fresh. Backup your specs, delete .Claude folder, then reinitialize in your project folder by following the instructions in the readme, it looks like your agent is doing legacy stuff hence it's failing. Using-agents is not a thing anymore, neither is the orchestrator.

This project was made too over complicated as I didn't really think through long term the impact of some feature requests I got previously.

This project may also no longer continue as I am working on an MCP server for this framework / workflow that works across all ai agents and isn't just tied to Claude code

[glebov]

Wow! MCP is perfect! I'm waiting it!

[glebov]

I noticed, that sub agent executer freezes sometimes, so he standing without actions 500ms and more.
Today i reached my limit, but did only 2 tasks, and yesterday i did 10 same tasks (without sub agents).