agentleFS
ToddGBensonToddGBenson/keel/.claude/skills/exploratory-testing/SKILL.md

exploratory-testing

Skill0 starsChanged 3 months agoApache-2.0View on GitHub

Run charter-driven exploratory testing to find defects scripted tests cannot reach. Use during the G4 QA gate, when asked to explore or manually test a feature, when a story is complex or risky, after an incident, or when scripted tests all pass but confidence is low. Covers charter design and the productive heuristics.

13 other projects use a version of this.

Adopted by

Most recently taken up by

Used with

Projects with it also use analytics-tracking-testing, compliance-testing and 5 more.

Other files people use alongside this one.

Files found in the same public projects as a version of this one, more often than they turn up anywhere else. Each project counts once, large collections count for less, and this file's own repository is left out.
  • petrkindlmannSkill

    analytics-tracking-testing

    Validate that analytics and marketing tracking fire CORRECTLY: GA4/GTM dataLayer events, Meta/TikTok/LinkedIn pixels, and ad-tech tags. Covers building a tracking plan as the contract, intercepting collect-endpoint beacons and dataLayer.push in Playwright, asserting event name + params + values + timing + de-duplication, Consent Mode v2 gating, CI regression gating, and news-media events (article-view, scroll-depth, paywall). Use when: "test analytics tracking," "GA4 event test," "verify the pixel fires," "dataLayer test," "tracking plan," "Meta Pixel dedup," "scroll-depth tracking test," "gate tracking in CI." Not for: whether tracking is ALLOWED to fire under consent law (GDPR/CMP) — that is compliance-testing; this skill checks the data is CORRECT. SEO meta tags / structured data — out of scope. Related: compliance-testing, playwright-automation, api-testing, qa-project-context.

    Also in 3 projects that use this

  • petrkindlmannSkill

    compliance-testing

    Test for regulatory compliance: GDPR/CMP consent verification, Google Consent Mode v2, Global Privacy Control (GPC), CCPA/US state opt-out, EU AI Act Article 50 transparency, Better Ads Standards, and cookie-inventory auditing. Covers automated consent-flow testing, policy-based script blocking before consent, and cookie drift detection. Use when: "GDPR test," "compliance," "CMP test," "cookie consent," "consent mode," "CCPA," "GPC," "AI Act," "Better Ads," "privacy banner." Not for: WCAG/axe-core test authoring — use accessibility-testing. Not for: OWASP/vuln scanning — use security-testing. Not for: evaluating your LLM feature's quality or safety — use ai-system-testing. Related: accessibility-testing, security-testing, ai-system-testing, ci-cd-integration.

    Also in 3 projects that use this

  • StephanSergioSkill

    test-automation

    Set up and write automated tests for a vibe-stack app — Vitest + React Testing Library for unit/component, Playwright for E2E, with Firebase and Claude mocked. Use when adding tests, a test harness, or a CI test gate.

    Also in 3 projects that use this

  • conorluddySkill

    accessibility-testing

    WCAG compliance testing and accessibility quality assurance workflows for iOS apps. Use when validating accessibility labels, testing VoiceOver compatibility, checking contrast ratios, or ensuring WCAG 2.1 compliance. Covers accessibility tree analysis, semantic validation, and automated accessibility testing patterns.

    Also in 3 projects that use this

  • +3 more from petrkindlmann

The file

We split this skill into the rules it gives an agent

4 rules. 0 are common, 4 are distinctive, and 0 are very specific to this project.

  • 0

    Common

    Most of the 11 edited versions we read keep these.

  • 4

    Distinctive

    This author's own choices. Worth a look.

  • 0

    Very specific

    Tied to their project's names, files or tools. Probably skip.

  • Common
  • Distinctive
  • Very specific
  1. Distinctive:

    Keep it to 45–90 minutes. Beyond that, attention degrades and note quality collapses — take a break and write a second charter instead.

    line 37
  2. Distinctive:

    Use it after having already used the adjacent feature.

    line 68
  3. Distinctive:

    Not afterward. Reconstruction is where the truth quietly changes, and the detail that turns out to matter is the one you did not think …

    line 75
  4. Distinctive:

    When the box ends, stop. Write the debrief.

    line 116

“Common” is measured: kept by at least 6 in 10 of the 11 edited versions of this file we could read. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.

Keep what fits and drop the rest. Your version, ready for your agents.
Files
SKILL.md

Exploratory testing

Consumers: qa-engineer, /qa-gate, process/gates/g4-verified.md.

Why it is not redundant with the test suite

Scripted tests find what someone already thought of. They encode yesterday's understanding, which is exactly the understanding that produced the bug.

Exploration is simultaneous learning, test design, and execution. It is where the expensive defects are found, because expensive defects live in the space nobody modelled — and it is the cheapest testing you can do per defect found.

It is not "clicking around". It is disciplined, charter-driven, time-boxed, and it produces evidence.

The charter

Written before you start. Without one, exploration drifts to whatever is most familiar and you re-test the happy path with extra steps.

Charter:   Explore draft recovery across concurrent sessions
           to discover data-loss and cross-user-visibility defects
Time-box:  45 minutes
Setup:     Two browsers, same user; a second user account; staging with synthetic data
Out:       Performance, accessibility (covered separately)

Format: Explore <area> with <resources> to discover <information>.

Keep it to 45–90 minutes. Beyond that, attention degrades and note quality collapses — take a break and write a second charter instead.

One risk per charter. A charter naming three unrelated risks will get one of them tested properly.

Heuristics that reliably find things

Concurrency and interleaving. Two tabs, same user, divergent edits. Two users, same resource. Submit twice quickly. Start an action, then start another before it completes. Race the save against the timeout.

State transitions. Log out mid-action. Let the session expire during an in-flight request. Change a permission while a session holds the old one. Navigate back after submitting. Refresh at every step. Use the browser's back button — it is the most under-tested control in every web application.

Boundaries, and just past them. Empty, one, exactly the limit, limit+1, maximum, maximum+1. Zero, negative, very large. Empty string vs. whitespace vs. null.

Data that looks nothing like demo data. Unicode, emoji, RTL text, names with apostrophes, very long strings, HTML, SQL fragments, newlines in single-line fields, leading and trailing whitespace, a filename with a semicolon.

Dependency degradation. Not "down" — slow. Throttle the network. That path is almost never tested and behaves far worse than the outage path, which at least has an error handler.

The misunderstanding path. What does a user do who has misread the feature? Who expects it to work like the competitor's? Who is halfway through and interrupted?

The second time. Most features are tested once, on a clean account. Use it five times. Use it after having already used the adjacent feature. Accumulate state.

Interruption. Close the laptop. Lose connectivity mid-request. Kill the tab during a save.

Note-taking, as you go

Not afterward. Reconstruction is where the truth quietly changes, and the detail that turns out to matter is the one you did not think worth writing.

14:02 two tabs, same doc, edits in both → last-write-wins, no conflict surfaced,
      no indication the other edit existed. Reproducible 3/3. → #204 (Major)
14:11 session expiry during in-flight autosave → draft saved under expired session,
      invisible after re-auth. Data is in the DB but unreachable. → #205 (High)
14:20 5 MB limit: eviction fires but the notice renders behind the modal — cosmetic
14:31 tried to reproduce #205 with a slow network instead of expiry — did not repro

Record what you tried that found nothing. It is the only thing that distinguishes "we tested that and it held" from "we never got to it", and at G4 that distinction is the difference between coverage and a coverage claim.

Debrief

At the end of the time-box:

  1. What did you find? File each as an issue with reproduction steps.
  2. What did you cover? Areas touched, and how deeply.
  3. What did you not reach? Deferred, out of scope, or blocked.
  4. What is the next charter? Findings usually suggest one — a defect in one area is evidence about the class of thinking that produced it.

The charter and the notes go to evidence/<issue>/g4/qa/exploratory-charter.md. They are SA-11 evidence.

Handing off findings

Reproduction steps, expected vs. actual, severity, and how consistently it reproduces. "Intermittent 1 in 5" is a legitimate and important finding — it usually means a race, and a race that reproduces one time in five in a manual session reproduces constantly in production.

You do not fix what you find. File it and hand to developer — a tester who fixes becomes a tester who stops looking hard.

Time-boxing is the discipline

When the box ends, stop. Write the debrief. If the area deserves more, that is a new charter with a fresh question — not an extension that runs until attention fails.

Controls

SA-11 (developer testing and evaluation) · CA-2 · SI-2 (defects found feed remediation).

Download now (ZIP)

Does it work?

Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.

No reviews yet

People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.

  • It worked
  • Partly
  • It didn't
What did you use it for? e.g. Reviewing pull requests in a Rails app

Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.

Your trail

Files you open will collect here, so you can find them again.

Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.

Now do this for your own.

Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.