parallel-investigation

1# Parallel Investigation
2 
3Coordinate parallel investigation threads to explore multiple hypotheses simultaneously. Most effective for production incidents, performance regressions, or integration failures where the root cause is unclear.
4 
5## Core Principle
6 
7**When uncertain, explore multiple paths in parallel. Converge when evidence points to an answer.**
8 
9Parallel investigation reduces time-to-solution by eliminating serial bottlenecks.
10 
11## Investigation Structure
12 
13### Phase 1: Problem Decomposition
14 
15Break the problem into independent investigation threads:
16 
17```
18Problem: API responses are slow
19 
20Investigation Threads:
21├── Thread A: Database performance
22│   └── Check slow queries, indexes, connection pool
23├── Thread B: Application code
24│   └── Profile endpoint handlers, check for N+1
25├── Thread C: Infrastructure
26│   └── Check CPU, memory, network latency
27└── Thread D: External services
28    └── Check third-party API response times
29```
30 
31Each thread should be independent (no blocking dependencies), focused (clear scope), and time-boxed.
32 
33### Phase 2: Thread Assignment
34 
35Assign threads with clear ownership:
36 
37```markdown
38## Thread A: Database Performance
39**Investigator:** [Name/Agent A]
40**Duration:** 30 minutes
41**Scope:**
42- Query execution times
43- Index utilization
44- Connection pool metrics
45**Report Format:** Summary + evidence
46```
47 
48### Phase 3: Parallel Execution
49 
50Each thread follows this pattern:
51 
521. Gather evidence specific to thread scope
532. Document findings as you go
54…