Achieve 100% unit test coverage

[KRRT7]

Goal

Reach 100% unit test coverage incrementally, starting from the current baseline of 60% (34,694 statements, 13,771 missed).

Baseline (2026-04-23)

Coverage	Module	Stmts	Missed
0%	main.py	95	95
0%	tracer.py	268	268
4%	lsp/	681	652
15%	telemetry/	33	28
20%	cli_cmds/	2,782	2,238
20%	tracing/	748	599
25%	api/	487	365
25%	optimization/	496	370
38%	languages/ (base)	1,830	1,129
43%	benchmarking/	1,518	871
45%	picklepatch/	163	90
60%	verification/	1,813	727
62%	setup/	803	307
63%	result/	340	126
69%	discovery/	1,200	369
71%	code_utils/	2,634	760
71%	github/	34	10
71%	languages/javascript	5,797	1,681
73%	languages/java	7,467	2,053
78%	either.py	27	6
81%	languages/python	4,587	875
86%	models/	863	124

Plan

Phase 0 — Infrastructure

Set up pytest-cov in CI, configure .coveragerc, add a coverage gate that fails CI if coverage drops below the current baseline. Every PR must maintain or improve coverage.

Phase 1 — Zero/near-zero coverage (0–25%)

Modules with no or minimal tests. These are the highest-risk areas.

• main.py, tracer.py, telemetry/
• lsp/
• tracing/
• api/
• optimization/
• cli_cmds/

Phase 2 — Low coverage (38–63%)

Modules with some tests but significant gaps.

• languages/ (base: function_optimizer.py, base.py, registry.py)
• benchmarking/
• picklepatch/
• verification/
• setup/
• result/

Phase 3 — Medium coverage to 100% (69–86%)

Modules with decent coverage that need gap-filling.

• discovery/
• code_utils/
• languages/javascript
• languages/java
• languages/python
• models/
• github/, either.py

Principles

• Every PR should maintain or improve coverage — never decrease it
• Prioritize modules by risk: core pipeline (optimization, verification, benchmarking) before peripheral (lsp, telemetry)
• Unit tests should be fast and isolated — mock external services (AI API, GitHub API)
• Use the coverage report to identify untested branches, not just untested files