Hermes Agent skill distilled from Google Testing Blog / Testing on the Toilet.
This repository is structured as a methodology skill, not as a chronological article digest. The skill body teaches an AI how to decide what testing method to use; detailed method cards live under references/methodologies/. Source article indexes, chunk summaries, and raw article text are kept under references/provenance/ for audit and future updates.
SKILL.md— concise AI operating guide and method router.references/methodologies/— primary references: structured testing methodology by topic.references/provenance/article-index.md— fetched article URL/date/title index.references/provenance/chunk-summaries/— 41 chunk summaries covering the fetched corpus.references/provenance/raw-articles/— local text snapshots of fetched articles, retained for provenance and future reprocessing.references/provenance/source-policy.md— source/provenance and copyright policy.
| File | Topic |
|---|---|
00-methodology-map.md |
Reference map and usage policy. |
01-test-strategy-risk.md |
Risk-driven strategy, enough testing, layers, production feedback. |
02-test-design-diagnostics.md |
Test values, assertions, names, diagnostic failures. |
03-testability-code-design.md |
DI, seams, globals, test doubles, code health. |
04-flakiness-hermeticity-ci.md |
Flakes, hermeticity, synchronization, CI/presubmit, coverage. |
05-ui-e2e-web-mobile.md |
UI, E2E, Web, mobile, localization, acceptance tests. |
06-specialized-systems.md |
Performance, concurrency, security, fuzzing, ML, media, API/config matrices. |
07-review-collaboration-culture.md |
Code review, testing roles, culture, knowledge propagation. |
08-extraction-and-knowledge-workflow.md |
How to maintain/update this skill from new posts. |
- Source: https://testing.googleblog.com/
- Fetched article count: 403
- Date range: 2007-01-21 through 2026-06-04
- Workflow: fetch all posts, split into chunks, use subagents to summarize, classify into methodology references, keep provenance separate.
Copy this directory into a Hermes skills directory or load it as a skill package. Use SKILL.md first; load detailed method files only when the task needs them. Use provenance files only for tracing or updating the skill.