Sistema de Inteligência em Qualidade de Software: executa, analisa, prevê, recomenda e aprende. Memória local + Learning Hub.
Three tools with partial schemas and descriptions in Portuguese. read_file and detect_project have basic descriptions (50-80 chars) but lack detailed parameter documentation. web_eval_browser has a longer description (180+ chars) but all three tools lack output schema documentation. Parameter descriptions exist but are minimal. No error handling guidance, no recovery patterns, no security considerations documented. Tool names follow verb_noun pattern (read_, detect_, web_eval_) which is positive, but descriptions lack LLM-optimization depth (should be 50-200 chars with WHEN/WHY context). Schemas present but incomplete, encoding enum in read_file is good, but missing type declarations on other params and no documented return structures.
Analisa o projeto e identifica frameworks de teste, pastas, backend, frontend, ambiente (web/mobile) e hints para geração de testes.
Lê o conteúdo de QUALQUER arquivo do projeto por caminho. Use para specs, page objects, componentes, código fonte - qualquer formato.
[Agente especializado: Browser] Abre a URL no navegador, captura screenshot, erros de console e requisições de rede. Inspirado em web-eval-agent. Requer: npm install playwright
No output schemas documented for any tool. LLMs cannot plan downstream calls or extract required fields without knowing return structure.
detect_project has empty input schema (no parameters documented). Unclear what analysis it performs or what context it needs.
Parameter descriptions are minimal (10-30 chars). 'Caminho relativo ao projeto' lacks format constraints, encoding details, or error guidance. Should specify: allowed file types, max path length, behavior on missing files.
No error handling guidance. What happens if read_file hits a permission error, missing file, or encoding mismatch? LLM has no recovery path.
| Scored | Grade | Overall | Spec posture | Rubric |
|---|---|---|---|---|
| 2026-09-23 | F | 46 | 2025-06-18+ | v2 |
web_eval_browser requires optional Playwright dependency but no fallback or clear error message if not installed. Tool descriptions should state prerequisites explicitly.