✨ OPRUN-4240: Add Claude Code integration for test profiling

.claude/commands/test-profile.md

@@ -0,0 +1,98 @@
+---
+description: Profile memory and CPU usage during e2e tests and analyze results
+---
+
+# Test Profiling
+
+Profile memory and CPU usage during e2e tests using the Go-based `bin/test-profile` tool.
+
+## Available Commands
+
+### /test-profile start [name]
+Start background profiling daemon. Collects heap/CPU profiles from operator-controller and catalogd every 10s.
+- Auto-stops after 3 consecutive collection failures (e.g., cluster teardown)
+- Use with any test command: `make test-e2e`, `make test-experimental-e2e`, etc.
+- Follow with `/test-profile stop` to generate analysis
+
+**Example:**
+```bash
+/test-profile start baseline
+make test-e2e
+/test-profile stop
+```
+
+### /test-profile stop
+Stop profiling daemon and generate analysis report. Cleans up port-forwards and empty profiles.
+
+### /test-profile run [name] [test-target]
+Automated workflow: start test, profile until completion, analyze.
+- Default test-target: `test-e2e`
+- Other targets: `test-experimental-e2e`, `test-upgrade-e2e`, etc.
+
+**Example:**
+```bash
+/test-profile run baseline test-e2e
+```
+
+### /test-profile analyze [name]
+Analyze existing profiles in `test-profiles/[name]/`. Generates markdown report with:
+- Memory/CPU growth patterns
+- Top allocators
+- OpenAPI, JSON, and cache hotspots
+
+### /test-profile compare [baseline] [optimized]
+Compare two test runs. Outputs to `test-profiles/comparisons/[baseline]-vs-[optimized].md`
+
+### /test-profile collect
+One-time snapshot of heap/CPU profiles from running pods. Saves to `test-profiles/manual/`
+
+## Implementation Steps
+
+When executing commands, I will:
+
+1. **Build tool**: `make build-test-profiler` (builds to `bin/test-profile`)
+2. **Execute command**: `./bin/test-profile [command] [args]`
+3. **For start/stop workflow**: Monitor daemon logs, handle errors gracefully
+4. **For run command**: Start test in background, monitor progress, analyze on completion
+5. **For analysis**: Present key findings from generated markdown reports
+
+## Configuration
+
+Environment variables (defaults shown):
+```bash
+TEST_PROFILE_COMPONENTS="operator-controller:olmv1-system:operator-controller-controller-manager:6060;catalogd:olmv1-system:catalogd-controller-manager:6060"
+TEST_PROFILE_INTERVAL=10           # seconds between collections
+TEST_PROFILE_CPU_DURATION=10       # CPU profiling duration
+TEST_PROFILE_MODE=both             # both|heap|cpu
+TEST_PROFILE_DIR=./test-profiles
+TEST_PROFILE_TEST_TARGET=test-e2e
+```
+
+## Output Structure
+
+```
+test-profiles/
+├── [name]/
+│   ├── operator-controller/
+│   │   ├── heap*.pprof
+│   │   └── cpu*.pprof
+│   ├── catalogd/
+│   │   ├── heap*.pprof
+│   │   └── cpu*.pprof
+│   ├── profiler.log
+│   └── analysis.md
+└── comparisons/
+    └── [name1]-vs-[name2].md
+```
+
+## Tool Location
+
+- Source: `hack/tools/test-profiling/` (Go-based CLI)
+- Binary: `bin/test-profile`
+- Make targets: `make build-test-profiler`, `make start-profiling/[name]`, `make stop-profiling`
+
+## Requirements
+
+- kubectl access to cluster
+- go tool pprof (for analysis)
+- Go version minimum from `hack/tools/test-profiling/go.mod` (for building)

.gitignore

@@ -38,8 +38,13 @@ vendor/
 \#*\#
 .\#*
 
-# AI temp files files
-.claude/
+# AI temp/local files
+.claude/settings.local.json
+.claude/history/
+.claude/cache/
+.claude/logs/
+.claude/.session*
+.claude/*.log
 
 # documentation website asset folder
 site