Ask JuhJuh

Type @juhjuh in any comment. JuhJuh's AI responds, grounded in your codebase, your knowledge base, and the surrounding conversation. No tab-switching. No copy-pasting into a separate chat window. The answer shows up right where the work happens.

How it works¶

1. Write a comment on a ticket or page
2. Include @juhjuh followed by your question or request
3. JuhJuh processes the mention in the background
4. A reply appears in the comment thread with the AI's response

The response draws from:

• The current ticket or page context: title, description, comments, and attached files
• Your project's codebase and file structure: the AI has read your repository
• Your organization's knowledge base (Pages): documented decisions, patterns, and guides
• Execution history and past decisions: what worked, what failed, and why

Where you can use @juhjuh¶

• Ticket comments: ask for implementation advice, clarify requirements, or get a code suggestion before execution
• Page comments: ask questions about documentation, request summaries, or find related pages
• Works anywhere comments are supported in JuhJuh

What to ask¶

You can ask anything that benefits from your project's context. Some examples:

• @juhjuh what files would this ticket affect?
• @juhjuh summarize the last three executions on this project
• @juhjuh how does our auth middleware work?
• @juhjuh suggest an approach for implementing this feature
• @juhjuh are there any related pages in our knowledge base?
• @juhjuh review this code snippet and flag potential issues

The more specific your question, the more useful the response. "@juhjuh how does our auth middleware handle token refresh?" will always beat "@juhjuh explain auth."

Context-aware responses¶

@juhjuh does not operate in a vacuum. Every response is shaped by layers of context that cascade together:

• Knowledge hierarchy: organization-level instructions, project-level conventions, team preferences, and user-specific context all feed into every answer. Higher-level rules take precedence, so your team's standards are always respected.
• Codebase awareness: the AI knows your repository's file structure, dependencies, import patterns, and architectural decisions. It references real files, not generic examples.
• Execution history: past successes and failures inform current suggestions. If a similar approach was tried and reverted last sprint, @juhjuh knows.
• Page confidence: your knowledge base pages carry a confidence score. Higher-confidence pages carry more weight in responses. Well-maintained documentation produces better answers.

Deduplication¶

JuhJuh processes each @mention exactly once. If a comment is edited after the AI has already responded, JuhJuh will not reprocess the same mention. Create a new comment for follow-up questions. This keeps threads clean and prevents duplicate responses.

Tips for better answers¶

• Be specific: include file names, function names, or ticket references when your question relates to them. Precision in, precision out.
• Provide context: if you are weighing two approaches, describe both. @juhjuh can compare trade-offs when it knows the options.
• Ask before execution: use @juhjuh to validate your approach before kicking off an execution. A quick sanity check saves time and compute.
• Combine with Pages: when @juhjuh gives you a useful answer, turn it into a Page. Your future self and your teammates will benefit.
• Chain questions: each follow-up comment can reference the previous answer. @juhjuh sees the full thread, so the conversation builds on itself naturally.

One mention, full context¶

Most AI assistants make you paste code into a separate window, explain the surrounding context, and hope for a relevant answer. @juhjuh already knows your codebase, your team's conventions, and the ticket you are looking at. You just ask the question.

That's the juju.

Related¶

• Pages: build the knowledge base that powers @juhjuh's answers
• Tickets: where most @juhjuh conversations happen
• Core Concepts: understand the knowledge hierarchy that shapes every response
• Agents: create specialized AI agents for targeted work beyond @mentions