name	analyze
description	Deep analysis mode - thorough multi-phase investigation with expert consultation

DEEP ANALYSIS MODE

Current Time: !date

Perform a comprehensive analysis using multi-phase expert consultation and parallel research.

Analysis Protocol

Phase 1: Reconnaissance (Parallel Swarm)

Launch immediately:

background_task(agent="explore", prompt="Map relevant code structure...")
background_task(agent="explore", prompt="Find related patterns...")
background_task(agent="librarian", prompt="Research best practices for...")

While swarm runs, use direct tools:

glob to find relevant files
grep to identify patterns
read to understand key code

Phase 2: Expert Analysis

Based on findings, consult specialists:

Domain	Agent	Focus
Architecture	`@architect`	System design, data flow, dependencies
Security	`@security`	Vulnerabilities, threat model, risks
Reliability	`@sre`	Scalability, SLOs, observability
Performance	`@perf`	Profiling, bottlenecks, optimization
Patterns	`@principal`	Code quality, best practices, tradeoffs

Phase 3: Deep Dive

Analyze comprehensively:

Edge cases and potential failure modes
Performance implications under load
Security attack surface
Error handling completeness
Testability and maintainability
Technical debt assessment

Phase 4: Synthesis

Combine findings into actionable output:

## Executive Summary

[2-3 sentences: key findings and recommendation]

## Detailed Analysis

### Architecture

[Component analysis, data flow, dependencies]

### Code Quality

[Patterns, anti-patterns, maintainability]

### Security

[Vulnerabilities, risks, compliance]

### Performance

[Bottlenecks, scalability concerns]

## Issues Found

### Critical (P0)

- [Issue with location and impact]

### High (P1)

- [Issue]

### Medium (P2)

- [Issue]

## Recommendations

### Immediate Actions

1. [Action with rationale]

### Short-term Improvements

1. [Improvement]

### Long-term Considerations

1. [Consideration]

## Trade-offs

[Analysis of different approaches with pros/cons]

Analysis Focus Areas

Area	What to Examine
Correctness	Logic errors, edge cases, assumptions
Security	Input validation, auth, data protection
Performance	Complexity, caching, resource usage
Reliability	Error handling, failure modes, recovery
Maintainability	Readability, coupling, documentation
Testability	Coverage, mocking, isolation

Output

Write to Obsidian via obsidian_append_content at: $OBSIDIAN_PATH/Analysis/YYYY-MM-DD-target.md

Note: $OBSIDIAN_PATH must be a vault-relative path (e.g., Projects/myapp), set per-project via direnv. The obsidian_append_content tool expects paths relative to the vault root.

Document Structure

Use this template for the Obsidian document:

@~/.config/opencode/templates/analysis-report.md

Analyze

$ARGUMENTS

analyze

Install Skill

SKILL.md