[nightshift] test-gap: test coverage gap analysis

[nightshift-micr]

Test Coverage Gap Analysis — Microck/kagi-cli

Summary

Module	Lines (approx)	Test Count	Coverage Assessment
src/api.rs	~5165	54	Good — extensive unit tests for parsing, normalization, stream handling
src/auth.rs	~880	19	Good — credential management, normalization, save/load
src/auth_wizard.rs	~655	9	Moderate — helper functions tested, interactive wizard untested
src/cli.rs	~1324	6	Adequate — validation and parsing tested
src/error.rs	~25	0	No tests — thin wrapper, low risk
src/http.rs	~169	2	Minimal — URL building tested, error mapping untested
src/main.rs	~1595	15	Moderate — output formatting tested, routing/command wiring untested
src/parser.rs	~917	12	Good — all major parsers tested
src/quick.rs	~756	10	Good — stream parsing and formatting tested
src/search.rs	~653	11	Good — validation, query building, error handling tested
src/types.rs	~902	0	No tests — data types only
tests/integration-cli.rs	~422	7	Sparse — only search, batch, auth check, summarize, news tested

Total: ~145 tests across 10 source modules + 1 integration test file.

---

P0 Critical — Core Business Logic Without ANY Tests

1. Integration tests missing for 8+ CLI commands

Commands with zero integration test coverage:

• kagi assistant (all subcommands: thread list/get/delete/export, custom assistant CRUD, prompt)
• kagi ask-page
• kagi quick
• kagi translate
• kagi fastgpt
• kagi enrich (web/news)
• kagi smallweb
• kagi lens (all subcommands)
• kagi bang (all subcommands)
• kagi redirect (all subcommands)

Only 5 of 13+ commands have integration tests.

2. execute_translate() orchestration untested

• File: src/api.rs lines ~1800–2800
• Individual parsing functions are unit-tested, but the full orchestration flow is not.
• Suggested test: Integration test using httpmock to mock translate endpoints.

3. Full CLI command routing untested

• File: src/main.rs lines 90–658
• The run() function dispatches to 13+ command variants with mostly untested paths.

---

P1 High — Error Handling Paths Untested

4. map_transport_error() — no unit tests

• File: src/http.rs line 33
• Suggested test: Verify timeout, connect, and generic error mappings.

5. HTTP status error branches in decode_kagi_*() functions untested

• File: src/api.rs lines 3508–3629
• None of the error branches (401, 403, 5xx, unexpected status) are tested.
• Suggested test: ~10 tests using mock reqwest::Response objects.

6. Search fallback logic (API → session token) untested

• File: src/main.rs lines 713–742
• Suggested test: Integration test where first API search returns 403, fallback succeeds.

7. run() error paths untested

• File: src/main.rs
• Including: bare auth non-interactive, conflicting flags, missing command, summarize flag combinations.

8. HTTP error status branches in search.rs and quick.rs

• Only #[ignore] live tests exist for error paths.

---

P2 Medium — Edge Cases Not Covered

9. Serde roundtrip tests missing

• File: src/types.rs (902 lines, 0 tests)
• Suggested test: Roundtrip tests for SearchResponse, AssistantPromptResponse, TranslateTextResponse, NewsStoriesResponse.

10. Parser error path tests with malformed HTML

• File: src/parser.rs
• All tests use valid HTML. No tests for missing title, missing href, missing attributes.

11. Batch search output formats untested

• File: src/main.rs lines 1129–1241
• Only JSON format integration-tested. Pretty, markdown, CSV formats untested.

12. Assistant stream error parsing

• Files: src/api.rs stream parsers
• Not tested with empty streams, unknown tags, or invalid JSON payloads.

---

P3 Low — Nice-to-Have

13. Display/format trait tests for CLI enums

14. format_status() with empty credential inventory

15. format_pretty_response() with published dates

16. format_client_error_suffix() with JSON/HTML/empty bodies

---

Recommended Actions

Priority	Action	Est. Tests
P0	Integration tests for 8+ CLI commands	~15-20 tests
P0	Unit tests for execute_translate() with mocks	~5 tests
P1	Tests for map_transport_error()	~3 tests
P1	HTTP status error branch tests	~10 tests
P1	Search fallback integration test	~1 test
P1	run() error path tests	~5 tests
P2	Serde roundtrip tests	~8 tests
P2	Parser error path tests	~6 tests
P2	Batch search format tests	~3 tests

---

Automated by Nightshift v3 (GLM 5.1) — merge if useful, close if not.