discover

<Use_When>

• User wants to identify improvement opportunities: /discover, "scan for improvements", "find issues"
• Starting work on an unfamiliar codebase and wanting a quality baseline
• Preparing for a release and wanting a comprehensive quality check
• User wants a prioritized backlog of technical debt </Use_When>

<Do_Not_Use_When>

• User wants to review a specific PR or changeset — use /deep-review instead
• User wants to fix a specific bug — delegate to debugger agent
• User wants to plan a feature — use /plan instead </Do_Not_Use_When>

<Why_This_Exists> Codebases accumulate technical debt across many dimensions — security gaps, missing tests, performance hotspots, documentation gaps, architectural drift. No single agent can efficiently cover all dimensions. Discover parallelizes specialist scans and produces a unified, deduplicated, severity-prioritized backlog that teams can act on. </Why_This_Exists>

<Execution_Policy>

• All 6 scan agents MUST run in parallel (independent concerns)
• Each agent focuses exclusively on its category — no overlap
• Consolidation agent runs after all scans complete
• Default scope: encourage explicit scoping to a subdirectory. Full-codebase scan only if user explicitly requests it
• Output: .omg/discover/backlog.md </Execution_Policy>

1. Launch 6 parallel scans: Fire all agents simultaneously

   Scan 1 — Security:

   Task(subagent_type="oh-my-copilot:security-reviewer", model="sonnet", name="discover-security", prompt="
   DISCOVERY SCAN: Security
   Scan the codebase at [SCOPE] for security vulnerabilities and risks.
   Focus on: OWASP Top 10, hardcoded secrets, injection risks, auth gaps, input validation.
   
   For each finding, output:
   FINDING:
   - category: security
   - severity: CRITICAL|HIGH|MEDIUM|LOW
   - title: [brief title]
   - file: [file:line]
   - description: [what's wrong and why it matters]
   - suggestedAction: [specific fix recommendation]
   END_FINDING
   ")
   

   Scan 2 — Quality:

   Task(subagent_type="oh-my-copilot:code-reviewer", model="sonnet", name="discover-quality", prompt="
   DISCOVERY SCAN: Code Quality
   Scan the codebase at [SCOPE] for anti-patterns, complexity hotspots, and SOLID violations.
   Focus on: God objects, high cyclomatic complexity (>10), deep nesting (>4 levels), code duplication, magic numbers.
   
   For each finding, output:
   FINDING:
   - category: quality
   - severity: CRITICAL|HIGH|MEDIUM|LOW
   - title: [brief title]
   - file: [file:line]
   - description: [what's wrong and why it matters]
   - suggestedAction: [specific fix recommendation]
   END_FINDING
   ")
   

   Scan 3 — Tests:

   Task(subagent_type="oh-my-copilot:test-engineer", model="sonnet", name="discover-tests", prompt="
   DISCOVERY SCAN: Test Coverage
   Scan the codebase at [SCOPE] for test coverage gaps and test quality issues.
   Focus on: untested public APIs, missing edge case tests, flaky test patterns, missing integration tests.
   
   For each finding, output:
   FINDING:
   - category: tests
   - severity: CRITICAL|HIGH|MEDIUM|LOW
   - title: [brief title]
   - file: [file:line]
   - description: [what's missing and why it matters]
   - suggestedAction: [specific test to add]
   END_FINDING
   ")
   

   Scan 4 — Performance:

   Task(subagent_type="oh-my-copilot:code-reviewer", model="sonnet", name="discover-perf", prompt="
   DISCOVERY SCAN: Performance
   Scan the codebase at [SCOPE] for performance issues and optimization opportunities.
   Focus on: N+1 queries, O(n²) algorithms, unnecessary re-renders, missing caching, synchronous I/O in hot paths, memory leaks.
   
   For each finding, output:
   FINDING:
   - category: performance
   - severity: CRITICAL|HIGH|MEDIUM|LOW
   - title: [brief title]
   - file: [file:line]
   - description: [what's slow and why it matters]
   - suggestedAction: [specific optimization]
   END_FINDING
   ")