coverage-report

Name: coverage-report
Author: livlign

18stars1forks

How to install coverage-report?

livlign/claude-skills/coverage-report

$npx -y skills add livlign/claude-skills --skill coverage-report

Installs into the current project.

Summary

Produce the latest deterministic C0/C1 coverage report for a .NET service. Use when the user asks for a 'coverage report', 'coverage status', 'what's covered', 'C0 C1 numbers', or wants to check the gate. Runs dotnet-coverage + ReportGenerator (numbers are tool output, never esti

Files · 1

View on GitHub

SKILL.md

1	# coverage-report
2
3	Numbers come only from the tool. The model never estimates, infers, or rounds coverage.
4	The model's only role is joining tool output to the manifest's reasons and narrating the
5	observations section.
6
7	Read `${CLAUDE_PLUGIN_ROOT}/rules/coverage-report.base.md` and the repo's
8	`.claude/coverage/refs/coverage-manifest.yml` first.
9
10	## One command, any time
11
12	The whole report is one wrapper — no arguments needed (it auto-detects the `.sln`):
13
14	```
15	./.claude/coverage/tools/report.sh
16	```
17
18	It collects coverage, joins it against the manifest, and **writes a dated per-run folder
19	`.claude/coverage/reports/<YYYY-MM-DD>/` containing `REPORT.md`, `REPORT.html`, and
20	`CANNOT-TEST.md`**, then prints the Markdown and exits non-zero if the ratchet fails (so it
21	doubles as a local gate). Each run writes its own dated folder rather than overwriting a fixed
22	path, so prior dates are preserved for comparison; override the date with `REPORT_DATE=YYYY-MM-DD`
23	to re-emit under a specific day. `CANNOT-TEST.md` is generated from the manifest `cannot_test`
24	entries (grouped by blocking construct, each row citing target, reason, and unlock), not
25	hand-written. This is the identical join CI runs — the report a developer sees locally is the report on
26	the PR. When invoked as the `coverage-report` skill, just run it and surface the result;
27	don't re-derive numbers by hand.
28
29	## Steps
30
31	1. Run the wrapper. `./.claude/coverage/tools/report.sh` (or, before init has copied it,
32	`${CLAUDE_PLUGIN_ROOT}/scripts/report.sh`). Pass a solution path or repo-filter only if
33	auto-detect is wrong. If the tools or solution are missing, report that — never fabricate
34	numbers. All numbers come from the tool + the deterministic join in `coverage-gate.py`; the
35	model does not parse XML or compute percentages.
36
37	2. Append the one narrated section — Observations. The script's report is complete and
38	deterministic (verdict, headline, Do-next, insights, by-bucket). The only thing the model
39	adds is Observations: suspected-latent-bug entries surfaced during generation (frozen,
40	not endorsed). Append them under an `## Observations` heading; add nothing else.
41
42	3. At baseline only. Write the measured Adjusted overall into `baseline.recorded_overall`
43	(with `basis: in-scope`) so it becomes the ratchet floor. Not on ordinary report runs.
44
45	## Report shape — the Unit Test Report (produced by `coverage-gate.py`)
46
47	Fixed six-section report card. Header carries commit / branch / date / tooling / verdict;
48	coverage is two-pass throughout — Adjusted (target set, the headline + ratchet basis) and
49	Raw (all instrumented), for C0, C1, and Method.
50
51	```
52	# Unit Test Report — <repo>
53	<commit · branch · date · tooling · verdict (gate PASS/FAIL · Adjusted C0/C1)>
54
55	## 1. Test Results total / passed / failed / skipped / flaky / duration (from .trx)
56	## 2. Coverage Summary Raw \| Adjusted \| Baseline \| Δ, for C0 / C1 / Method; + the Gate block
57	(ratchet always; diff-coverage + scope-change guard in PR mode, --base)
58	## 3. Coverage by Layer columns: Layer \| Raw lines \| Raw branches \| Testable lines \|
59	Testable branches \| C0 \| C1 — where C0/C1 cells are "NN% (covered/total)"
60	over the testable slice (covered counts fold into the cell, no separate
61	covered columns). Rows are
62	human LAYER names (Application/Service — TARGET, Infrastructure/Integration,
63	Presentation/Workers, Models/DTOs, Generated, Non-product) mapped from the
64	manifest categories — NEVER the raw `excl: <category>` strings (an `excl:` row
65	showing a coverage % read as "excluded but covered??"). Raw = all code in the
66	layer; Testable = its unit-testable slice (whole file in the target layer; only
67	carve-out methods elsewhere); C0/C1 are of the testable slice. Only the target
68	layer feeds the Adjusted headline. + a per-file table (same columns + the file's
69	Layer) so a 1631-line infra file reading as 131 testable lines is clear.
70	## 4. Risk Hotspots methods with complexity ≥5 and coverage <80%, sorted TARGET-FIRST, with a
71	"What to do" column + per-bucket guidance (target=unit-test; integration=
72	integration test/extract carve-out; e2e=E2E/refactor) and a takeaway line
73	calling out how many are real in-

People who liked this also viewed

6 picks

Type

coverage-report

18stars1forks

View on GitHub

How to install coverage-report?

livlign/claude-skills/coverage-report

$npx -y skills add livlign/claude-skills --skill coverage-report

Installs into the current project.

Summary

Files · 1

View on GitHub

SKILL.md

1	# coverage-report
2
3	Numbers come only from the tool. The model never estimates, infers, or rounds coverage.
4	The model's only role is joining tool output to the manifest's reasons and narrating the
5	observations section.
6
7	Read `${CLAUDE_PLUGIN_ROOT}/rules/coverage-report.base.md` and the repo's
8	`.claude/coverage/refs/coverage-manifest.yml` first.
9
10	## One command, any time
11
12	The whole report is one wrapper — no arguments needed (it auto-detects the `.sln`):
13
14	```
15	./.claude/coverage/tools/report.sh
16	```
17
18	It collects coverage, joins it against the manifest, and **writes a dated per-run folder
19	`.claude/coverage/reports/<YYYY-MM-DD>/` containing `REPORT.md`, `REPORT.html`, and
20	`CANNOT-TEST.md`**, then prints the Markdown and exits non-zero if the ratchet fails (so it
21	doubles as a local gate). Each run writes its own dated folder rather than overwriting a fixed
22	path, so prior dates are preserved for comparison; override the date with `REPORT_DATE=YYYY-MM-DD`
23	to re-emit under a specific day. `CANNOT-TEST.md` is generated from the manifest `cannot_test`
24	entries (grouped by blocking construct, each row citing target, reason, and unlock), not
25	hand-written. This is the identical join CI runs — the report a developer sees locally is the report on
26	the PR. When invoked as the `coverage-report` skill, just run it and surface the result;
27	don't re-derive numbers by hand.
28
29	## Steps
30
31	1. Run the wrapper. `./.claude/coverage/tools/report.sh` (or, before init has copied it,
32	`${CLAUDE_PLUGIN_ROOT}/scripts/report.sh`). Pass a solution path or repo-filter only if
33	auto-detect is wrong. If the tools or solution are missing, report that — never fabricate
34	numbers. All numbers come from the tool + the deterministic join in `coverage-gate.py`; the
35	model does not parse XML or compute percentages.
36
37	2. Append the one narrated section — Observations. The script's report is complete and
38	deterministic (verdict, headline, Do-next, insights, by-bucket). The only thing the model
39	adds is Observations: suspected-latent-bug entries surfaced during generation (frozen,
40	not endorsed). Append them under an `## Observations` heading; add nothing else.
41
42	3. At baseline only. Write the measured Adjusted overall into `baseline.recorded_overall`
43	(with `basis: in-scope`) so it becomes the ratchet floor. Not on ordinary report runs.
44
45	## Report shape — the Unit Test Report (produced by `coverage-gate.py`)
46
47	Fixed six-section report card. Header carries commit / branch / date / tooling / verdict;
48	coverage is two-pass throughout — Adjusted (target set, the headline + ratchet basis) and
49	Raw (all instrumented), for C0, C1, and Method.
50
51	```
52	# Unit Test Report — <repo>
53	<commit · branch · date · tooling · verdict (gate PASS/FAIL · Adjusted C0/C1)>
54
55	## 1. Test Results total / passed / failed / skipped / flaky / duration (from .trx)
56	## 2. Coverage Summary Raw \| Adjusted \| Baseline \| Δ, for C0 / C1 / Method; + the Gate block
57	(ratchet always; diff-coverage + scope-change guard in PR mode, --base)
58	## 3. Coverage by Layer columns: Layer \| Raw lines \| Raw branches \| Testable lines \|
59	Testable branches \| C0 \| C1 — where C0/C1 cells are "NN% (covered/total)"
60	over the testable slice (covered counts fold into the cell, no separate
61	covered columns). Rows are
62	human LAYER names (Application/Service — TARGET, Infrastructure/Integration,
63	Presentation/Workers, Models/DTOs, Generated, Non-product) mapped from the
64	manifest categories — NEVER the raw `excl: <category>` strings (an `excl:` row
65	showing a coverage % read as "excluded but covered??"). Raw = all code in the
66	layer; Testable = its unit-testable slice (whole file in the target layer; only
67	carve-out methods elsewhere); C0/C1 are of the testable slice. Only the target
68	layer feeds the Adjusted headline. + a per-file table (same columns + the file's
69	Layer) so a 1631-line infra file reading as 131 testable lines is clear.
70	## 4. Risk Hotspots methods with complexity ≥5 and coverage <80%, sorted TARGET-FIRST, with a
71	"What to do" column + per-bucket guidance (target=unit-test; integration=
72	integration test/extract carve-out; e2e=E2E/refactor) and a takeaway line
73	calling out how many are real in-

People who liked this also viewed

6 picks

Type