# tests/
This directory contains all functional tests for the `claude_runner_core` crate, which provides builder pattern API for executing Claude Code commands programmatically with zero process execution logic.
## Responsibility Table
| `builder_edge_cases_test.rs` | Test builder pattern edge cases |
| `builder_methods_test.rs` | Test builder method existence and chainability |
| `default_values_test.rs` | Test default value correctness |
| `environment_variables_test.rs` | Test environment variable setting |
| `migration_validation_test.rs` | Test migration completeness |
| `old_way_impossible_verification.rs` | Test deprecated pattern prevention |
| `responsibility_single_execution_point_test.rs` | Test single execution point principle |
| `types_test.rs` | Test type definitions and conversions |
| `verification_impossibility_test.rs` | Test impossibility layer verification (VER-3) |
| `verification_migration_metrics_test.rs` | Test migration metrics verification (VER-2) |
| `verification_negative_criteria_test.rs` | Test negative criteria verification (VER-6) |
| `verification_rollback_test.rs` | Test rollback impossibility verification (VER-4) |
| `verification_shortcuts_test.rs` | Test shortcuts detection verification (VER-5) |
| `manual_edge_case_check.rs` | Test edge case manual verification |
| `float_edge_cases_test.rs` | Test float parameter edge cases (NaN, infinity, negative) |
| `verbose_and_path_edge_cases_test.rs` | Test verbose flag and path parameter edge cases |
| `describe_test.rs` | Test describe() and describe_env() command inspection methods |
| `execution_output_test.rs` | Test ExecutionOutput struct fields and Display formatting |
| `skip_permissions_test.rs` | Test skip_permissions flag and --dangerously-skip-permissions arg |
| `session_dir_tests.rs` | Test SessionManager directory creation and Strategy string parsing |
| `process_test.rs` | Process scanner: `/proc` scan, signal sending (TC-061–TC-070) |
| `dry_run_test.rs` | Test dry_run mode and describe_compact() output (TSK-071) |
| `io_params_test.rs` | Test I/O parameter builder methods (TSK-072) |
| `tool_dir_params_test.rs` | Test tool and directory parameter builder methods (TSK-073) |
| `session_params_test.rs` | Test session parameter builder methods (TSK-074) |
| `prompt_permission_params_test.rs` | Test prompt and permission parameter builder methods (TSK-075) |
| `model_budget_params_test.rs` | Test model and budget parameter builder methods (TSK-076) |
| `mcp_extension_params_test.rs` | Test MCP and extension parameter builder methods (TSK-077) |
| `debug_advanced_params_test.rs` | Test debug and advanced parameter builder methods (TSK-078) |
| `terminal_ide_params_test.rs` | Test terminal and IDE parameter builder methods (TSK-079) |
| `pattern_e_empty_and_edge_cases_test.rs` | Test Pattern E empty-iterator bug fix and float edge cases |
## Organization (31 test files)
Tests organized by functional domain and architectural principles (see Responsibility Table above).
### Scope
This test suite covers the claude_runner_core crate's builder pattern API for Claude Code command construction and comprehensive verification framework (31 test files):
**In Scope:**
- Builder pattern API (4 test files):
- Edge cases: token limits (0, 1, 200K, u32::MAX), method overrides (last wins), argument accumulation
- Methods: 61+ with_*() methods, chainability, order independence
- Defaults: tier 1 (bash_timeout=3.6M, auto_continue=true, telemetry=false, max_output_tokens=200K, chrome=--chrome), tier 2/3 (None)
- Environment variables: each parameter sets correct env var, tier 1 defaults set vars
- Type definitions (1 test file):
- ActionMode enum (Ask/Allow/Deny) with string conversions
- LogLevel enum (Error/Warn/Info/Debug/Trace) with ordering and conversions
- Migration validation (2 test files):
- Factory pattern completely removed, builder pattern universally adopted
- Old patterns impossible (from_message/create/generate dont exist)
- Single execution point enforcement
- Verification framework - 6-layer pyramid (5 test files, 231 validations total):
- Layer 1 - Migration metrics (42 checks): 0%→100% shift in 8 metrics
- Layer 2 - Rollback detection (27 checks): migration irreversible
- Layer 3 - Impossibility (34 checks): old API wont compile
- Layer 4 - Shortcuts detection (48 checks): no mocks/fakes/disabled tests
- Layer 5 - Negative criteria (15 checks): forbidden patterns = 0
- Layer 6 - Positive tests (65 tests): new pattern works
- Process scanning and signal delivery (1 test file):
- `/proc` enumeration: returns Vec without panic, excludes self PID, finds spawned processes, handles deleted CWD
- Signal sending: `send_sigterm` / `send_sigkill` with valid and invalid PIDs (TC-061–TC-070)
- Bug reproducers with 5-section documentation:
- issue-token-limit-default: Migration from factory pattern didnt preserve 200K default
- Root Cause, Why Not Caught Initially, Fix Applied, Prevention, Pitfall documented
- Test organization: Test Matrix cataloging, Lessons Learned documentation, Cross-references in Out of Scope
**Out of Scope:**
- Session lifecycle management (→ claude_profile crate)
- Context injection from wplan (→ dream_agent crate)
- Actual process execution (→ dream_agent crate)
- Interactive terminal UI (→ terminal-based tools)
- Configuration hierarchy (→ config_hierarchy crate)
**Test Quality**: Uses process::Command inspection via test-only `build_command_for_test()` method. All tests validate builder output without executing Claude binary. Token limit tests validate environment variables. Override tests validate last-wins semantics. Accumulation tests validate args collection. Bug reproducer tests include complete 5-section documentation following test_organization.rulebook.md standards.
## Navigation
Test files follow domain-based naming that reflects the functionality being tested. Bug reproducer tests use `_bug_` suffix with issue ID. Edge case tests use descriptive suffixes (`_edge_case`, `_minimum`, `_boundary`, `_stress`).