feat(0.2): finish Track 3 centerpiece — integration / e2e / unified PR comment

[pmclSF]

Summary

The three Track 3 deliverables that close the pitch's "maps how
your unit, integration, e2e, and AI tests actually relate to your
code" promise. Track 3 is the centerpiece of the parity-gated
0.2.0 plan; this PR finishes the three remaining items (3.3–3.5).

• Track 3.3 — Integration-test classification rigor.
  Content-based detection of HTTP-testing libraries (supertest,
  httptest, MockMvc, requests, etc.) so integration tests living
  alongside unit tests in flat directories are correctly classified.
• Track 3.4 — E2E-to-code attribution honest carve-out.
  Documents the exact limit of e2e attribution in 0.2 — structural-
  only, with the deeper coverage-instrumented path explicitly out
  of scope until 0.3.
• Track 3.5 — Unified PR-comment rendering audit. A render-
  golden test enforces the visual contract: bracketed [LABEL]
  badges, **\path`**` locators, em-dash separators, and a single
  recommended-tests stanza for unit + integration + e2e.

What changed

New code (internal/testtype/, internal/analysis/):

• integration_imports.go — pattern allowlist (30+ patterns
  across JS/TS, Go, Python, Java, Ruby) + InferFromContent and
  MergeContentInference helpers
• integration_classification.go — analyzer wiring that reads each
  test file once via the existing FileCache and refines the
  test-type classification

New tests:

• integration_imports_test.go — 14 tests covering positive
  matches, prose-mention false-positive guards, multi-library
  co-detection, and merge logic
• integration_classification_test.go — 5 tests covering the
  promote-to-integration path, the leave-pure-unit-alone guard,
  the Go httptest case, the nil-cache path, and cancellation
• unified_render_test.go — render-golden enforcing the four
  uniformity gates plus consistent section ordering

New docs (docs/product/):

• test-type-classification.md — what 0.2 detects, what it
  doesn't, false-positive catalog, extension recipe
• e2e-attribution.md — the honest carve-out (structural-only)
  and the 0.3 roadmap
• unified-pr-comment.md — the visual contract, the alternatives
  considered, the AI section-level grouping exception

Why these three together

They're independent in code but co-located in the plan: all three
are "the pitch's promise about test types is now defensible."
Bundling minimizes review thrash on tightly-related work.

Test plan

• go build ./... clean
• go test ./internal/... all pass (incl. 19 new tests)
• go test ./cmd/... all pass
• go test ./internal/testdata/... all pass
• make docs-verify passes
• Manual smoke: terrain analyze on a repo with integration
  tests in flat directories → integration classification fires
• Manual smoke: terrain report pr --format markdown on a
  mixed unit/integration/e2e + AI PR → unified comment shape

Plan tracker

Closes Track 3.3, 3.4, 3.5 from the parity-gated 0.2.0 plan.
Track 3 (the centerpiece) is now complete: 3.1 (PR --fail-on)
and 3.2 (--explain-selection) shipped earlier in the cycle; 3.3
through 3.5 land here. Remaining: Tracks 5, 6, 7, 9, 10.2-10.8.

🤖 Generated with Claude Code

[github-actions]

Terrain AI Risk Review

Metric	Value
AI surfaces	13
Eval scenarios	16
Impacted scenarios	0
Uncovered surfaces	13

Decision: PASS — AI surfaces are covered.

[github-actions]

[RISK] Terrain — Merge with caution

High-severity gaps found in changed code.

Metric	Value
Changed files	9 (3 source · 3 test)
Impacted units	7
Protection gaps	3
Tests selected	11 of 775 (1% of suite)

---

Coverage gaps in changed code

• internal/analysis/analyzer.go [MED] — Exported function Analyze has no observed test coverage.
  → Add unit tests for exported function Analyze — this is public API surface.
• internal/analysis/analyzer.go [MED] — Exported function AnalyzeContext has no observed test coverage.
  → Add unit tests for exported function AnalyzeContext — this is public API surface.
• internal/analysis/integration_classification.go [LOW] — integration_classification.go has no observed test coverage.
  → Add unit tests for integration_classification.go.

4 pre-existing issues on changed files

• internal/analysis/analyzer.go [HIGH] — [blastRadiusHotspot] Changes to this file propagate to 954 tests (450 direct, 504 indirect). High blast radius increases regression risk.
• internal/analysis/integration_classification.go [HIGH] — [blastRadiusHotspot] Changes to this file propagate to 954 tests (450 direct, 504 indirect). High blast radius increases regression risk.
• internal/testtype/integration_imports.go [HIGH] — [blastRadiusHotspot] Changes to this file propagate to 1578 tests (32 direct, 1546 indirect). High blast radius increases regression risk.
• internal/analysis/integration_classification_test.go [LOW] — [fixtureFragilityHotspot] Fixture 'mustWrite' is used by 5 tests across 1 files. A single change cascades widely.

---

Recommended tests

11 test(s) with exact coverage of 4 impacted unit(s). 3 impacted unit(s) have no covering tests in the selected set.

Test	Confidence	Why
internal/analysis/analyzer_test.go	exact	exact coverage of New
internal/analysis/bench_test.go	exact	exact coverage of New
internal/analysis/context_test.go	exact	exact coverage of New
internal/analysis/fixture_deps_test.go	exact	exact coverage of Analyzer, New
internal/analysis/gitignore_test.go	exact	exact coverage of Analyzer, New
internal/analysis/integration_classification_test.go	exact	test file directly changed
internal/analysis/tsconfig_extends_test.go	exact	exact coverage of Analyzer, New
internal/analysis/vitest_in_source_test.go	exact	exact coverage of Analyzer, New
internal/changescope/unified_render_test.go	exact	test file directly changed
internal/engine/pipeline_test.go	exact	exact coverage of New
internal/testtype/integration_imports_test.go	exact	exact coverage of InferFromContent, MergeContentInference

---

Owners: PMCLSF

Limitations

• No coverage artifacts provided; protection gaps reflect missing data, not measured absence. Provide --coverage to improve accuracy.
• Mixed test cultures reduce cross-framework optimization confidence. Consider standardizing on fewer frameworks.

_{Generated by Terrain · terrain pr --json for machine-readable output}

Targeted Test Results

Terrain selected 11 test(s) instead of the full suite.

• Go tests: passed