Skip to content
SkillModel-led4,094 tokens in use

test-harness

Rules for tests where a green gate means something: what each harness fakes, how to run and report the gate, regression tests for fixes, and lessons from real bugs.

680 stars on the repository MIT Licensephuryn/product-engineering · ai-shipping/skills/test-harness
claude plugin install ai-shipping@product-engineering
View source

Install command as written in the repository's README

Skill label

measured from its files

What installing test-harness costs

43tokens idle
4,094tokens in use
21,069tokens, up to

Idle is what sits in context while the skill is installed. In use is its instructions, loaded when it runs. Up to adds the files its instructions link. Counts are approximate, made offline with cl100k_base.

Model dependence Model-led

Script-ledMixedModel-led

Mostly instructions, so how well it works depends on the model that follows them.

This describes the skill's structure, how much of it is scripts. It does not say how well it works.

Touches

shelldetectedgitdetected

Declared tags come from the skill's frontmatter; detected ones from its scripts, its code blocks and the commands it names in inline code. This is detected, not audited.

How it scores

Quality

How well the skill is written, counted from its own files: its frontmatter, its steps, examples, stated output and constraints, its size and whether every file it ships is one it names.How we score it →
Good
  • Frontmatter8/15
  • Structure17/25
  • Examples3/15
  • Output15/15
  • Size11/15
  • Resources15/15

Security

Starts at 100 and loses points for each thing in the skill's files that a reader should look at first.How we score it →
Excellent

No findings.

Repository health

Pre-computed score based on development activity, maintenance, community, maturity, and trend momentum.How we score it →
Good
  • Development activity100
  • Maintenance100
  • Community44
  • Maturity4
  • Momentum40

test-harness starts from one question: what does each harness fake? A bug survives where every suite fakes the same side, so the skill lays out unit, integration, schema-catalog, browser, visual, live-agent and smoke harnesses, what each makes real, and where each runs.

It defines the local gate, how to read a run (the exit code, never the printed counts), what to do when two fixes pass their checks and leave the symptom unchanged, and a list of rules learned from bugs.

What You Get

  • A harness table that says what each suite makes real, what it fakes and whether it runs in the gate
  • A gate definition: typecheck, test, end-to-end and authenticated end-to-end in order, with visual, live-agent and smoke runs kept outside on purpose
  • Rules for regression tests, flaky runs, hangs and fixes that do not land, with a stop-and-observe rule after two failed fixes
  • A boundary-test list: cross-tenant refusal, once-only actions, script tags shown as text, replayed work acting once, and every page rendering for every role
  • Reference files for harnesses, captures, vendor surfaces, external creates and the coverage map

Common Use Cases

  • Deciding which suite should cover a new feature and where it runs
  • Proving a bug fix by running its test against the broken behavior first
  • Investigating a red, flaky or hanging run
  • Writing a coverage map that records the runs that happened

What Makes It Unique

It is organized around what each test layer fakes and carries concrete rules such as ‘a fallback in a test is a bug in disguise’ and ‘the gate is the exit code’. It limits who may write the coverage map to whoever ran the gate, requires naming every suite that did not run in a preview or go message, and refuses sleeps, retries and skips as cures for flakes.

Under The Hood

The skill is markdown only: one main file, a receipts file and six reference files, and it ships no scripts. The agent it guides runs the product’s own test, build and browser commands, starts local servers and development databases, and writes test files, coverage-map entries and capture output. It tells the agent to commit through a heredoc and to read exit codes from logs. It never tests against production data, and the smoke check is the only run that touches production, without writing.

Join founders building with open source

Opinionated takes, migration guides, cost-saving tips, and insights from the open source ecosystem.

Subscribe on SubstackJoin 750+ subscribers