deleting-thoughts

Guards irreversible delete_thought MCP calls with id resolution, confirmation, and derivative checks.

4.5k|872|Updated Mar 11, 2026
One-click install
npx skills add https://github.com/NateBJones-Projects/OB1 --skill deleting-thoughts
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: deleting-thoughts
Source: https://github.com/NateBJones-Projects/OB1/tree/main/skills/deleting-thoughts
Command: npx skills add https://github.com/NateBJones-Projects/OB1 --skill deleting-thoughts

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

The delete_thought MCP tool performs a hard, irreversible delete with no undo, trash, or restore, so a single wrong call permanently destroys a stored memory. This Skill enforces a safe deletion workflow so AI clients never delete an unverified or wrong thought.

Core Features & Use Cases

  • Verified ID Resolution: Requires resolving the target UUID via search_thoughts or list_thoughts in the current session before any delete call.
  • Confirmation Gate: Surfaces the thought's content to the user and demands explicit confirmation before firing delete_thought.
  • Derivative Protection: Runs find_derivatives or related_thoughts first to avoid orphaning provenance chains, steering outdated thoughts toward updating or tagging as superseded instead.
  • Use Case: When you ask your AI client to "delete my test thought about the Q3 roadmap", the client searches for the thought, shows you its content, checks for derived thoughts, and only deletes after you confirm — then reports the deletion receipt.

Quick Start

Ask your AI client to delete a specific Open Brain thought and watch it search, show the content, and request confirmation before deleting.

Frequently Asked Questions about deleting-thoughts

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I safely delete an Open Brain thought with MCP?

Resolve the thought's UUID using search_thoughts or list_thoughts in the current session, show its content to the user, get explicit confirmation, then call delete_thought. Never delete an id typed from memory or guessed.

Can I undo a delete_thought call in Open Brain?

No, delete_thought is a hard delete with no undo, trash, or restore. Recovery depends entirely on database backups, which is why confirmation before deletion is mandatory.

When should I update instead of delete a thought?

Update or tag a thought as superseded when it is merely outdated, wrong, or replaced by newer information. Delete only test thoughts, exact duplicates, or accidental captures where editing is not sufficient.

What happens to derived thoughts when I delete their source?

Deleting a source thought orphans the provenance chain of any thoughts derived from it. Run find_derivatives or related_thoughts first, and prefer deprecating over deleting when derivatives exist.

Why does delete_thought return Thought not found?

The id was already deleted or incorrect. Re-resolve the target via search_thoughts rather than retrying with a guessed UUID, since a wrong guess could delete the wrong memory permanently.

Which AI clients support this deletion safety skill?

Claude Code, Codex, Grok, and any AI client supporting reusable skills or custom instructions connected to the open-brain-delete-thought MCP server. Only the skills directory path differs per client.