- Home
- Skills
- Code Quality & Review
- Diff Coverage Checker: Tests for Every Changed Line
More screenshots
Works with the AI Agents you already use
Diff Coverage Checker: Tests for Every Changed Line
Turn any coverage report into the only number that matters for a PR: which changed lines no test runs, grouped by file, with a CI gate.
$9
About this skill
"Coverage is 87%" — and none of the code the agent just wrote is tested. The total barely moves when new code ships untested, so nobody notices.
Measured on pallets/click: an untested helper plus a new module no test imports dropped the project number from 87.2% to 86.9%. Diff Coverage Gate failed the change at 5.2% (1 of 19 changed lines) and listed both ranges.
Diff Coverage Gate gives your agent one script that answers "is my change tested?" in about a second:
- Reads the report you already have: Cobertura XML (pytest-cov, coverlet, gcovr), JaCoCo XML (Maven, Gradle), lcov (Jest, c8, Node, cargo-llvm-cov), Go coverprofile, coverage.py JSON, Istanbul JSON. Auto-detected and auto-discovered.
- Intersects it with your git diff from the merge-base, including staged, unstaged and new untracked files (the ones agents create and never
git add). - Lists every uncovered changed range per file, worst first, with the code, so the agent knows exactly which tests to write.
- Gates: exit 1 under
--min(default 80%). JSON output and a GitHub Actions job included.
Benchmark on real repositories (pinned commits, commands in the bundle):
| Repo | Report | Clean range | Seeded untested change |
|---|---|---|---|
| pallets/click | Cobertura / JSON / lcov | 92.0% (185 of 201), same in all 3 formats | FAIL 5.2% |
| tj/commander.js | lcov | 100%, 0 false positives on 4 ranges | FAIL 25.0% |
| apache/commons-cli | JaCoCo | 100% (39 of 39) | FAIL 0.0% |
| go-chi/chi | Go coverprofile | 100% (109 of 109) | FAIL 0.0% |
On click and commons-cli the line-by-line result matches an established open-source diff-coverage tool exactly. Every run took 0.7-1.7 s. The first commander.js run flagged config and example files as untested; that was fixed and is now covered by a regression test (54 tests in total).
What it does not do: it does not run your tests or produce coverage, and it trusts the report (if your coverage tool counts a comment line, so does the gate). It measures executed lines, not assertion quality. Partial branches are shown, not gated.
Works in Claude Code, Cursor, Codex CLI, Copilot and any SKILL.md agent.
Private by design: one Python file, standard library only, read-only git commands, no network, no telemetry.
How to use
- 1
Connect your agent to the Agensi MCP
Copy the prompt and paste it into Claude, ChatGPT, Cursor or any agent that supports MCP. It sets up the connection for you. Manual setup
- 2
Sign in to Agensi
Your agent shows a sign-in link. Sign in and click Allow.
- 3
Ask for what you need
Your agent loads this skill and uses it on your task. It always asks before any payment.
Reviews
No reviews yet
Be one of the first to try it. Every listed skill goes through the trust checks below.
Security scanned
Checked by our 8-point scan before listing
Fresh listing
Recently published to Agensi
Money-back guarantee
Doesn't do what its page promises? You get your money back.
Trust & safety
Security scan
Security scan passed 27 days ago
Requires