Deep Analysis Skill

This skill grants access to the sequentialthinking tool. Use this when you need to "break down" a problem before acting.

When to use

• Debugging a complex race condition or "impossible" bug.
• Planning a large-scale refactor.
• Analyzing security implications of a change.
• When the user asks "What do you think?" or "Analyze this".

Available Tools (Context Loaded)

• mcp_sequential-thinking_sequentialthinking: A tool that allows you to have an internal monologue/thought process. You can update your hypothesis, revise thoughts, and branch your thinking.

Best Practices

1. Don't rush: Use multiple thought steps (totalThoughts > 5) for hard problems.
2. Revise: If a thought turns out to be wrong, use the isRevision flag to correct yourself.
3. Hypothesis: Always state a working hypothesis early in the thought chain.

Example Workflow

1. User: "Why is the auth service crashing?"
2. Agent: Calls sequentialthinking with thought="Initial hypothesis: DB connection pool exhaustion...".
3. Agent: Calls sequentialthinking with thought="Checking logs..." and needsMoreThoughts=true.
4. ... (After 5-10 thoughts) ...
5. Agent: "The issue is a deadlock in the transaction manager..."