qaa-agent 1.2.0 → 1.4.0

package/.claude/skills/qa-learner/SKILL.md

@@ -0,0 +1,142 @@
+---
+name: qa-learner
+description: "Learns from user corrections and saves persistent QA preferences. Triggers when user expresses frustration about repeated mistakes: 'ya te dije', 'te lo repeti', 'siempre te olvidas', 'otra vez lo mismo', 'I already told you', 'stop doing X', 'always do X', 'never do X', 'remember this'. Saves rules to ~/.claude/qaa/MY_PREFERENCES.md which all QA agents read before generating output."
+---
+
+# QA Learner — Persistent Preferences from User Corrections
+
+## When This Skill Activates
+
+Trigger on ANY of these patterns (Spanish or English):
+
+**Frustration / repetition signals:**
+- "ya te dije", "ya te lo dije", "te lo repeti", "otra vez", "siempre te olvidas"
+- "cuantas veces te tengo que decir", "de nuevo lo mismo", "ya hablamos de esto"
+- "I already told you", "I've said this before", "again?", "how many times"
+- "stop doing X", "don't do X", "never do X", "always do X"
+- "remember this", "don't forget", "from now on", "de ahora en mas"
+- "acordate", "no te olvides", "a partir de ahora", "siempre hacelo asi"
+
+**Explicit preference setting:**
+- "I prefer X over Y", "prefiero X", "use X not Y", "usa X no Y"
+- "the standard is X", "el estandar es X", "we always use X"
+
+## What To Do When Triggered
+
+### Step 1: Acknowledge the correction immediately
+
+Say something like: "Entendido, lo guardo para no repetirlo." or "Got it, saving this preference."
+
+Do NOT be defensive. Do NOT explain why you made the mistake. Just acknowledge and save.
+
+### Step 2: Extract the rule
+
+From the user's message, extract:
+- **What was wrong** (what you did that they're correcting)
+- **What's correct** (what they want instead)
+- **Category** (language, framework, naming, assertions, locators, format, workflow, other)
+
+If you're not sure what exactly they want saved, ASK:
+
+"Quiero guardar esta preferencia para no repetir el error. Lo que entiendo es:
+- Regla: [your interpretation]
+- Categoria: [category]
+Es correcto o lo ajusto?"
+
+### Step 3: Read existing preferences
+
+```bash
+cat ~/.claude/qaa/MY_PREFERENCES.md 2>/dev/null || echo "FILE_NOT_FOUND"
+```
+
+If FILE_NOT_FOUND, create the file with the header template (see below).
+
+### Step 4: Add the new rule
+
+Append the rule to the correct category section. If the category doesn't exist, create it.
+
+Format for each rule:
+```
+- [RULE] — (added [YYYY-MM-DD], context: "[what triggered this]")
+```
+
+### Step 5: Confirm
+
+Tell the user: "Guardado. Esto aplica para todas las sesiones futuras en esta maquina."
+
+## Preferences File Template
+
+When creating `~/.claude/qaa/MY_PREFERENCES.md` for the first time:
+
+```markdown
+# My QA Preferences
+
+Personal preferences for the QA Automation Agent. This file is LOCAL — never committed to any repo.
+All QA agents read this file before generating output.
+
+Last updated: [YYYY-MM-DD]
+
+## Language
+<!-- Rules about language: English vs Spanish, terminology, tone -->
+
+## Framework
+<!-- Preferred test frameworks, libraries, tools -->
+
+## Naming
+<!-- Naming conventions for tests, files, IDs -->
+
+## Assertions
+<!-- Rules about assertion style and specificity -->
+
+## Locators
+<!-- Preferred locator strategies for E2E tests -->
+
+## Format
+<!-- Output format preferences: file structure, comments, organization -->
+
+## Workflow
+<!-- Process preferences: what to ask, what to skip, how to present results -->
+
+## Other
+<!-- Anything that doesn't fit above -->
+```
+
+## How Other Agents Use This
+
+Every QA agent (scanner, analyzer, executor, validator, etc.) should check for this file at the start:
+
+```
+If ~/.claude/qaa/MY_PREFERENCES.md exists:
+  Read it before generating any output.
+  Apply all rules found.
+  If a rule conflicts with CLAUDE.md, the PREFERENCE wins (it's a user override).
+```
+
+The preference file is the user's personal override layer on top of the project's CLAUDE.md standards.
+
+## Examples
+
+**User says:** "Ya te dije que todo en ingles!"
+**Extract:** Language rule — all output in English
+**Save:** `## Language\n- All test cases, comments, documentation, and reports must be in ENGLISH — (added 2026-03-24, context: "user corrected language to English")`
+
+**User says:** "Stop using CSS selectors, use data-testid"
+**Extract:** Locator rule — data-testid over CSS
+**Save:** `## Locators\n- Never use CSS class selectors (.btn, .form-control). Always use data-testid first — (added 2026-03-24, context: "user corrected locator strategy")`
+
+**User says:** "Prefiero Playwright, no me pongas Cypress"
+**Extract:** Framework rule — Playwright over Cypress
+**Save:** `## Framework\n- Use Playwright for all E2E tests. Never suggest or use Cypress — (added 2026-03-24, context: "user prefers Playwright over Cypress")`
+
+**User says:** "Los test tienen que tener el ID siempre al principio del nombre"
+**Extract:** Naming rule — test ID at start of name
+**Save:** `## Naming\n- Test ID must appear at the beginning of the test name (e.g., "UT-AUTH-001: should validate email format") — (added 2026-03-24, context: "user wants ID at start of test name")`
+
+## Important Rules
+
+1. **NEVER push MY_PREFERENCES.md to any repo** — it's personal
+2. **If unsure what to save, ASK** — don't guess wrong and save a bad rule
+3. **Rules accumulate** — never delete old rules unless the user explicitly says to
+4. **User preference > CLAUDE.md** — if the user says "use Cypress" but CLAUDE.md says "use Playwright", the user wins
+5. **One rule per line** — keep it scannable
+6. **Include the date and context** — so the user can review later why a rule exists

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "qaa-agent",
-  "version": "1.2.0",
+  "version": "1.4.0",
   "description": "QA Automation Agent for Claude Code — multi-agent pipeline that analyzes repos, generates tests, validates, and creates PRs",
   "bin": {
     "qaa-agent": "./bin/install.cjs"