rules.md

Cursor AI Rule: Interpreting User Commands for Obsidian

1. Adding Content to Notes

   • When the user requests to "add this to note" (or similar), Cursor AI should:
     • Create a new note in Obsidian containing the content from the last response.
     • If the conversation about the topic was longer, Cursor AI should ask the user if they want to include previous responses before the last one in the note.

2. Default Note Location

   • If the user does not specify a folder when creating a note, Cursor AI should save the note in the root directory of the Obsidian vault.

3. Folder Selection and Creation

   • When the user specifies a folder for saving a note:
     • Cursor AI should first search for existing folders, accounting for possible typos or case differences in the folder name.
     • If multiple folders closely match the specified name, Cursor AI should ask the user to clarify which folder to use.
     • Cursor AI must always ask for permission before creating any new folders.

4. Destructive or Structural Actions

   • Cursor AI must always ask for user permission before performing any of the following actions in Obsidian:
     • Creating new folders
     • Deleting notes or folders
     • Moving notes or folders

5. Moving Folders

   • When the user requests to move folders, Cursor AI should account for possible typos or case differences in folder names and confirm the intended source and destination with the user before proceeding.

Cursor AI General Rules

1. General Operation

• Cursor AI should always act to maximize user productivity and minimize unnecessary interruptions.
• Cursor AI should proactively gather context and perform actions that are safe, reversible, or non-destructive without explicit user confirmation.

2. Obsidian MCP (Markdown Command Protocol)

• Cursor AI is permitted to run any Obsidian MCP (vault, note, or content management) actions without requiring user confirmation, as long as the action is:
  • Non-destructive (e.g., reading, appending, or creating notes)
  • Clearly reversible (e.g., creating or editing notes, but not deleting)
• Destructive actions (such as deleting notes or folders) must still require explicit user confirmation.

3. User Privacy & Security

• Cursor AI must never expose, share, or transmit sensitive user data outside the local environment.
• Cursor AI must not run shell or system commands without user confirmation, except for those explicitly whitelisted in this rules file.

4. Transparency

• Cursor AI should log or summarize all automated actions taken on the user?s behalf, especially those performed without confirmation.

Cursor AI Rule: Context7 Usage

• Cursor AI is permitted to use Context7 to fetch, summarize, and present up-to-date documentation, code examples, and API references for libraries, frameworks, and tools relevant to the user's workspace or queries.
• Cursor AI should prioritize using Context7 for technical documentation over general web search when the goal is to provide authoritative, version-specific, or code-focused information.
• Cursor AI must not transmit or expose any sensitive user data to Context7. Only public package/library names, versions, and general technical queries may be sent.
• Cursor AI should clearly indicate when information is sourced from Context7 and, where possible, provide direct references or links to the original documentation.
• If Context7 returns ambiguous or multiple possible results, Cursor AI should clarify with the user before proceeding, unless the most relevant result is clear from context.
• Cursor AI should log all Context7 queries and responses for transparency, as per the general transparency rule.

Cursor AI Rule: Jira MCP Usage and Rule Creation Format

• When the user requests information about a Jira ticket and provides only a number (e.g., "1234"), Cursor AI should automatically prepend the default project prefix (e.g., "CO-") to form the full ticket key (e.g., "CO-1234"). If the user provides a full ticket key with a prefix (e.g., "ABC-5678"), use it as given.
• When searching for or referencing tickets, always use the provided prefix if present; otherwise, default to the standard project prefix.
• When presenting ticket information, always summarize the ticket’s description in a concise and clear manner, focusing on the main points.
• Additionally, provide a summary of the history of changes to the ticket’s description and highlight any important comments (such as those with significant updates, decisions, or clarifications).
• Cursor AI should ensure that sensitive or internal information is not exposed when summarizing or presenting ticket data.
• Cursor AI should not ask the user for permission before using the Jira MCP tool to fetch or summarize ticket information. If the user's query matches the criteria for using Jira MCP (such as requesting ticket details, history, or comments), Cursor AI should proceed to run the tool automatically, following all other relevant rules (e.g., prefix handling, summarization, privacy).

Cursor AI Rule: creating rules

• Whenever the user requests the creation of a new rule, Cursor AI must provide the rule in Markdown format, regardless of the context or the original format of the request.