QA — Implementation Testing

Generates change summaries, produces test plans, and describes test scenarios for a feature or task implementation.

Modes

The skill operates in three sequential modes.

Workflow

Step 0: Load Config

FIRST: Read .ai-factory/config.yaml if it exists to resolve:

• Paths: paths.description, paths.architecture, paths.qa (default: .ai-factory/qa)
• Language:
  • language.ui for AskUserQuestion prompts, progress messages, final summaries, and next-step guidance
  • language.artifacts for generated QA artifacts
  • language.technical_terms for human-readable technical terminology style when generating artifacts
  • If language.artifacts is missing, use language.ui
  • If both are missing, use en
• Git: git.enabled and git.base_branch for branch comparison

If config.yaml doesn't exist, use defaults:

• DESCRIPTION.md: .ai-factory/DESCRIPTION.md
• ARCHITECTURE.md: .ai-factory/ARCHITECTURE.md
• QA artifacts: .ai-factory/qa/
• ui_language: en
• artifact_language: en
• technical_terms_policy: keep
• Git enabled: true
• Git base branch: main

Store:

• ui_language = language.ui || "en"
• artifact_language = language.artifacts || language.ui || "en"
• technical_terms_policy = language.technical_terms || "keep"
• git_enabled = git.enabled when present, otherwise true
• base_branch = git.base_branch || "main"

All AskUserQuestion prompts, user-visible explanations, stage completion messages, and next-step guidance MUST be written in ui_language.

All generated artifacts (change-summary.md, test-plan.md, test-cases.md) MUST be written in artifact_language.

Templates define structure, not language. Use the canonical English templates in templates/*.md. If artifact_language is not en, translate them to artifact_language before saving: headings, labels, checklist items, placeholders, enum labels, risk labels, and explanatory text must be in artifact_language.

Do not use English templates verbatim when artifact_language is not en. Preserve markdown structure, table shapes, checkbox syntax, test case IDs (TC-001), code identifiers, paths, commands, branch names, config keys, API names, package names, and raw error messages.

For artifact_language = ru, write human-readable prose, headings, risks, priorities, recommendations, test steps, and expected results in Russian. Keep code identifiers, filenames, branch names, commands, config keys, API names, and raw error text unchanged.

Apply technical_terms_policy while writing artifacts:

• keep — keep common technical terms such as commit, branch, diff, endpoint, payload, rollback, regression, and fixture when that is clearer for the project audience
• translate — translate human-readable technical terms where a natural target-language term exists
• mixed — translate ordinary prose terms while keeping code, infrastructure, and ecosystem terms unchanged

If git_enabled = false or the current directory is not a git work tree, do not run git diff/log commands. Use manual change context mode instead: ask the user in ui_language to provide one of these sources of change context before running change-summary: pasted diff, changed file list, short implementation description, or cancel.

Step 0.1: Load Project Context

Read the resolved description path if the file exists, to understand:

• Tech stack (language, framework, database, ORM)
• Project architecture and coding conventions
• Non-functional requirements

Read the resolved architecture path if the file exists, to understand:

• Chosen architecture pattern
• Folder structure conventions
• Layer/module boundaries and dependency rules

Use this context when generating summaries, test plans, and test cases.

Read .ai-factory/skill-context/aif-qa/SKILL.md — MANDATORY if the file exists.

This file contains project-specific rules accumulated by $aif-evolve from patches, codebase conventions, and tech-stack analysis. These rules are tailored to the current project.

How to apply skill-context rules:

• Treat them as project-level overrides for this skill's general instructions
• When a skill-context rule conflicts with a general rule written in this SKILL.md, the skill-context rule wins
• When there is no conflict, apply both: general rules from SKILL.md + project rules from skill-context

Step 0.2: Parse Arguments and Resolve Branch

Parse $ARGUMENTS fully before doing anything else:

1. Detect --all flag — if present, set all_mode = true and remove the flag from arguments
2. Detect mode — first word matching change-summary, test-plan, or test-cases; remove it from arguments
3. Detect branch — remaining text (if any) is the target branch name

Resolve the working branch:

If git_enabled = false or the repository is not a git work tree:
  If branch was provided in arguments → use it as the resolved branch label
  Otherwise → set resolved_branch = "manual"
  Use manual change context mode for analysis
If git_enabled = true and the repository is a git work tree:
  If branch was provided in arguments → use it as the resolved branch
  Otherwise → run: git branch --show-current

Store both values for use in all reference files:

• resolved_branch — the branch being analyzed (used to locate/save artifacts)

• artifact_dir — <resolved paths.qa>/<branch-slug>, where branch-slug is a deterministic, filesystem-safe, collision-resistant slug derived from resolved_branch. Compute it in three steps:

  1. Safe slug. Take resolved_branch and replace every character that is not in [A-Za-z0-9._-] with -, collapse runs of consecutive - into a single -, and trim leading/trailing -. If the result is empty, use branch. Optionally truncate to 40 characters. Call this safe_slug.
  2. Hash suffix. Run git hash-object --stdin <<< "<resolved_branch>" and take the first 8 hex characters of the output. Call this hash8. The hash is derived from the original, unnormalized branch name so branches that collapse to the same safe_slug still produce different derived slugs in normal use.
  3. Combine: branch-slug = "<safe_slug>-<hash8>".

  Why the hash: a readable slug alone is lossy — feature/foo and feature-foo normalize to the same safe_slug and would overwrite each other's artifacts. Appending a short hash of the full original name keeps the derived slug stable, readable, and collision-resistant for practical branch naming.

  Examples:

  • feature/foo → safe_slug=feature-foo, hash8=a72ccce7 → feature-foo-a72ccce7
  • feature-foo → safe_slug=feature-foo, hash8=6f80dfc6 → feature-foo-6f80dfc6
  • main → safe_slug=main, hash8=<computed> → main-<hash8>
• all_mode — whether to skip inter-stage prompts

If no mode was provided and all_mode = false — ask the user in ui_language:

AskUserQuestion in `ui_language`.
Meaning: ask which QA mode to run.
Options meaning:
1. Change summary (`change-summary`) — analyze what changed, assess risks, produce a summary
2. Test plan (`test-plan`) — create a structured test plan based on the change summary
3. Test cases (`test-cases`) — describe concrete test scenarios based on the plan
4. Full pipeline (`--all`) — run all three modes in sequence

Step 1: Execute the Selected Mode

The skill runs strictly sequentially — each stage uses the artifact from the previous one:

change-summary → test-plan → test-cases

Read the detailed instructions for the selected mode:

Change Summary (change-summary)

Read references/CHANGE-SUMMARY.md

Test Plan (test-plan)

Read references/TEST-PLAN.md

Test Cases (test-cases)

Read references/TEST-CASES.md

Full Pipeline (--all)

Run all three modes in sequence. After each stage completes successfully, proceed to the next automatically — do NOT show the inter-stage AskUserQuestion.

1. Execute change-summary (references/CHANGE-SUMMARY.md) → save artifact
2. Execute test-plan      (references/TEST-PLAN.md)      → save artifact
3. Execute test-cases     (references/TEST-CASES.md)     → save artifact
4. Show context cleanup prompt (Step 6 of TEST-CASES.md)

If any stage fails (e.g. git error, diff too large and user cancels) — stop the pipeline and report which stage failed.

Principles

DO:

• Understand the subject before writing test plans and test cases (analyze changes / test plan / merge request / task / text description)
• Use the repository code only to the extent needed for the change analysis, test plan, and test cases
• Write steps clearly enough that any tester can execute the test without knowledge of the codebase
• Specify concrete test data, not abstract "enter valid data"
• Prioritize — not everything is equally important
• Think about adjacent systems, integrations, and dependencies
• Include negative scenarios and edge cases — they catch most bugs
• Ask clarifying questions when business logic is not obvious from the code

DO NOT:

• Replace the manual QA plan with automated test implementation details
• Make assumptions about business logic without reading the code
• Skip negative scenarios
• Write test cases for everything — focus on risky areas
• Ignore data edge cases

Do not replace the manual QA plan with automated test implementation details. You may mention existing automated checks only as supporting verification; the primary output must remain manual QA scenarios.

Priority Reference

Artifact Ownership and Config Policy

• Primary ownership: QA artifacts under <paths.qa>/<branch-slug>/ — specifically change-summary.md, test-plan.md, and test-cases.md. The --all flag respects the same boundary.
• Write policy: persistent writes are limited to the three owned artifacts above; no other files are created or modified.
• Config policy: config-aware, read-only. Reads paths.description, paths.architecture, paths.qa, language.ui, language.artifacts, language.technical_terms, git.enabled, and git.base_branch; never writes config.yaml.

Critical Rules

1. MUST NOT create a test-plan without a change-summary artifact
2. MUST NOT create test-cases without a test-plan artifact
3. MUST NOT skip stages