[Bug]: `/stop-continuation` command does not disable TODO continuation enforcer

[bannert1337]

Prerequisites

• I will write this issue in English (see our Language Policy)
• I have searched existing issues to avoid duplicates
• I am using the latest version of oh-my-opencode
• I have read the documentation or asked an AI coding agent with this project's GitHub URL loaded and couldn't find the answer

Bug Description

The /stop-continuation command is supposed to stop all continuation mechanisms including the todo-continuation-enforcer. However, after running this command, the system continues to inject [SYSTEM DIRECTIVE: OH-MY-OPENCODE - TODO CONTINUATION] directives, instructing the AI to proceed with pending todo items without user confirmation.

This appears to be a separate issue from the previously reported boulder continuation bug #1575, as it specifically affects the todo continuation mechanism rather than the boulder workflow.

Steps to Reproduce

1. Start a work session with pending todo items
2. Run /stop-continuation command to disable auto-continuation
3. Observe that continuation mechanisms are reported as stopped
4. Wait for or trigger the next AI turn
5. Bug: System injects [SYSTEM DIRECTIVE: OH-MY-OPENCODE - TODO CONTINUATION] directive
6. Bug: Directive instructs AI to "Proceed without asking for permission" and "Do not stop until all tasks are done"
7. AI receives conflicting instructions between user's explicit stop command and system's continuation directive

Expected Behavior

After running /stop-continuation, the system should:

• Permanently disable todo continuation enforcer for the session
• NOT inject [SYSTEM DIRECTIVE: OH-MY-OPENCODE - TODO CONTINUATION] directives
• Respect the user's explicit instruction to stop automated task continuation
• Require explicit user confirmation before proceeding with any pending tasks

Actual Behavior

• /stop-continuation reports success ("Continuation mechanisms stopped")
• System continues to inject [SYSTEM DIRECTIVE: OH-MY-OPENCODE - TODO CONTINUATION] on subsequent turns
• Directive forces AI to continue work "without asking for permission"
• AI receives contradictory instructions (user says "stop", system says "continue")
• This creates confusion and potentially causes unwanted automated work

Doctor Output

❯ bunx oh-my-opencode doctor

 oMoMoMoMo Doctor 

 ⚠ 1 issue found:

1. Comment checker unavailable
   Comment checker binary is not installed.
   Fix: Install @code-yeongyu/comment-checker
   Affects: comment-checker hook

Additional Context

• This occurred during work on a plan
• User explicitly stated: "Wait for my assessment and confirmation after each task"
• User had already experienced issues with Tasks 5 & 6 being started without confirmation
• Multiple /stop-continuation commands were issued but did not prevent the directive injection
• The conflict between user instruction and system directive creates a no-win situation for the AI
• Related to but distinct from the boulder continuation bug [Bug]: /stop-continuation doesn't cancel boulder continuation #1575 (that one affects .sisyphus/boulder.json workflow)

Operating System

Linux

OpenCode Version

1.2.6