Test generation
Curated, auditable, benchmark-ready Agent Skills library for Claude Code, Codex, OpenCode, Cursor, and more.
npx -y skills add shinzoxD/knackbox --skill test-generationAssembled from the repository path, not quoted from the project. Check it against their README if it does not work.
2 things to look at
- 27 days oldThe repository was created 27 days ago. New is not bad, but a brand new repository carrying a familiar-sounding name is the shape a typosquat arrives in, and there has been no time for anyone else to find a problem with it.
- 0 stars0 stars. Stars are a popularity signal and not a quality one, but at this level it is likely that nobody has read this closely except its author, and you would be relying on your own review.
What its author says it does
Copied from the file, not written here
Design and write focused automated tests for new code, bug fixes, and risky behavior. Use whenever the user asks for tests, shares untested code, fixes a defect that needs regression coverage, mentions coverage gaps, or wants help deciding what to test, even if they do not name a testing framework.
The file declares its own license as Apache-2.0. That is the author’s claim about this one file, and it is not the same thing as the license GitHub reports for the repository, which is listed with the other numbers below.
SKILL.md
3.0 KB, as published. Nobody here has run it
Test Generation
Create tests that protect observable behavior and likely failure modes. A small set of high-signal tests is better than broad coverage coupled to implementation details.
Workflow
- Identify the public contract, invariants, side effects, and failure modes.
- Read the repository's existing test framework, fixtures, naming, and helper patterns before writing code.
- Rank cases by risk: regression, boundaries, permissions, malformed input, concurrency, and external failures.
- Choose the narrowest test level that proves the behavior.
- Write deterministic tests with an explicit arrange, act, assert structure.
- Run the smallest relevant test command, then the broader suite when the change affects shared behavior.
Case selection
Include cases from these groups when relevant:
- Representative success path.
- Boundary values immediately below, at, and above a limit.
- Empty, missing, malformed, duplicated, and unexpectedly large input.
- Authorization and tenant isolation.
- Dependency timeout, partial failure, retry, and cancellation.
- State transitions, repeated calls, and idempotency.
- A regression test that fails for the reported bug before the fix.
Output format
When code can be edited, provide the tests and summarize:
## Coverage added
- `<test name>`: behavior and risk protected
## Command
`<exact test command>`
## Remaining gaps
- Anything requiring integration infrastructure or unavailable context
When only planning tests, return a table with case, level, setup, assertion, and priority.
Rules
- Assert externally meaningful results, not private method call sequences.
- Do not use snapshots for volatile or security-sensitive values unless the snapshot is intentionally reviewed and stable.
- Avoid sleeps, network access, wall-clock assumptions, and random data without a fixed seed.
- Keep one reason for failure per test where practical.
- Reuse local factories and fixtures, but do not hide the behavior under test behind excessive helpers.
- Never claim a test passes unless it was run; report the exact blocker if it could not be executed.
Edge cases
- For legacy code with no test seam, first add a characterization test around current behavior.
- For generated code, test the generator or schema rather than generated lines.
- For flaky failures, preserve diagnostics and isolate nondeterminism before weakening assertions.