feat(testing): add MCP and CLI test harnesses

[galligan]

Add transport harnesses for integration testing:

• createMcpHarness(): In-memory MCP transport for testing
• createCliHarness(): CLI execution with exit code capture
• TestTransport: Bidirectional message passing
• Memory sink for capturing CLI output
• Comprehensive test coverage (29 tests)

Co-Authored-By: Claude Opus 4.5 noreply@anthropic.com

Resolves #47

[galligan]

Warning

This pull request is not mergeable via GitHub because a downstack PR is open. Once all requirements are satisfied, merge this PR as a stack on Graphite.
Learn more

• chore(outfitter): sync templates on pack #108
• chore(changesets): ignore app package #107
• fix(errors): avoid TaggedError extends warnings #106
• chore(release): align package versions to 0.1.0 #105
• docs: add README and CHANGELOG for testing, kit, outfitter #95
• feat(templates): add daemon project template #94
• feat(templates): add MCP server project template #93
• feat(templates): add CLI project template #92
• fix(cli): implement confirmDestructive prompts #104
• fix(cli): implement createCLI and command builder #103
• feat(kit): add @outfitter/kit meta-package #91
• docs(daemon): add README and CHANGELOG #90
• fix(daemon): fix locking module bugs (GREEN phase) #89
• test(daemon): add locking module tests (RED phase) #88
• feat(daemon): add errors, platform, and locking module stubs #87
• feat(outfitter): add doctor command for environment validation #86
• feat(outfitter): add init command with project scaffolding #85
• feat(daemon): add IPC communication and health checks #84
• feat(daemon): add types and lifecycle management #83
• feat(index): add FTS5 full-text search implementation #82
• feat(index): add types for FTS5 full-text search #81
• feat(testing): add MCP and CLI test harnesses #80 👈 (View in Graphite)
• feat(testing): add test fixture utilities #79
• feat(mcp): add MCP server implementation with typed handlers #78
• feat(mcp): add types and interfaces for MCP server #77
• feat(ui): add Shape discriminated union and render() #76
• feat(ui): add createTokens semantic color layer #75
• feat(state): add pagination helpers #74
• feat(logging): add DEFAULT_PATTERNS for common secret redaction #73
• feat(ui): add formatRelative() for human-readable timestamps #72
• feat(state): add encodeCursor/decodeCursor for pagination #71
• feat(types): add hashId() for deterministic short identifiers #70
• feat(contracts): add Logger.child() and fix AssertionError category #69
• feat(config): centralize environment variable access with Zod validation #42
• fix(logging): enable redaction by default when global rules exist #41
• docs(ui): add README, CHANGELOG, and TSDoc comments #39
• docs(state): add README, CHANGELOG, and TSDoc comments #38
• docs(file-ops): add README, CHANGELOG, and TSDoc comments #37
• docs(logging): add README, CHANGELOG, and TSDoc comments #36
• docs(config): add README, CHANGELOG, and TSDoc comments #35
• fix(cli): enforce cursor TTL #114
• fix(tests): remove placeholder tests #113
• chore(gitattributes): prefer ours for bun lockfiles #115
• main

This stack of pull requests is managed by Graphite. Learn more about stacking.

[greptile-apps]

Greptile Summary

• Implements comprehensive testing harnesses (createMcpHarness and createCliHarness) for transport-agnostic integration testing of MCP servers and CLI commands
• Adds fixtures module with utilities for test data factories, temporary directories, and environment variable isolation
• Creates complete test suite with 29 tests following the project's TDD-first approach where tests define expected behavior

Important Files Changed

Filename	Overview
packages/testing/src/index.ts	New entry point exporting all testing utilities with comprehensive JSDoc documentation
packages/testing/src/mcp-harness.ts	New MCP test harness providing tool invocation, search, and fixture loading capabilities
packages/testing/src/cli-harness.ts	New CLI test harness for executing commands and capturing stdout/stderr/exit codes

Confidence score: 4/5

• This PR is safe to merge with high confidence as it implements well-structured testing infrastructure
• Score reflects comprehensive test coverage and follows established patterns, but deducted one point due to test files being written before implementation (TDD approach means some tests may initially fail)
• Pay close attention to packages/testing/src/__tests__/fixtures.test.ts which references fixture files that may not exist yet

Sequence Diagram

sequenceDiagram
    participant User

Loading

[galligan]

Reviewed the CLI harness concerns: stdin is closed immediately and signal-based exits are mapped to 128+signal; tests passed in the pre-push run.

[galligan]

Verified CLI harness now closes stdin to allow EOF and maps signal termination to non-zero exit codes (128+signal). Restacked/checked and submitted with gt.

[galligan]

Restacked after downstack update (formatRelative test stabilization); no additional changes in this PR.

[galligan]

CLI harness already closes stdin immediately and maps signal terminations to non-zero (128+signal) exit codes. Resubmitted after restack.