Skip to content

CHANGELOG.md

Latest commit

 

History

History
101 lines (87 loc) · 5.38 KB

CHANGELOG.md

File metadata and controls

101 lines (87 loc) · 5.38 KB

Changelog

All notable changes to this project will be documented in this file.

[v5.0.0] - 2026-06-02

Added

  • Evo-12: MCP Expansion — 4 MCPs cientificos (latest-science, research-mcp, sura-papers, arxiv-mcp) = 10 fontes academicas unificadas
  • Evo-12: 28 skills de datasets (gnomAD, GTEx, dbSNP, Ensembl, PDB, STRING, PubChem, OpenFDA, etc.) — total 38 science skills
  • Evo-13: 4 Reasoning Engines — Z3 4.16 (prova formal), SymPy 1.14 (simbolico), miniKanren (logica relacional), Critical (15 falacias + vieses)
  • Evo-11: 9 skills cientificas core (AlphaFold, PubMed, ChEMBL, UniProt, FoldSeek, ClinVar, PyMOL, OpenAlex)
  • Evo-10: SandeClaw Integration — ProviderFactory (Multi-LLM fallback), Hot-Reload Skills, ReAct Agent Loop
  • Evo-10: 5 MCPs criticos restaurados + SciHub reabilitado
  • Evo-9: Qualis Target Navigator (7 fatores, 49 areas CAPES)

Changed

  • Skills: 46 → 150 (+104)
  • MCPs: 41 → 46 (+5)
  • Science skills: 2 → 38
  • Reasoning engines: 0 → 4
  • TDD suites: 226
  • SDD specs: 162
  • Science: 38
  • Agency agents: 26
  • MCP health: 58% → 50%
  • README, AGENTS.md, AGENTS_PTBR.md atualizados para v5.0.0

[v4.6.1] - 2026-05-27

Added

  • Taxonomia universal: 350 tipos de raciocinio em 35 categorias (I-XXXV), 335 referencias
  • 11 novas categorias: Quantico-Informacionais, Cibernetico-Organizacionais, Geometrico-Topologicos, Linguistico-Semanticos, Psicometrico-Avaliativos, Climatico-Ambientais, Etico-Normativos Avancados, Esportivo-Performaticos, Musical-Acusticos, Culinario-Gastronomicos, Metafisico-Ontologicos
  • 150 novos tipos de raciocinio (R201-R350)
  • Integracao multimodal visao+texto (skill multimodal-vision): Vision Router com 3 providers (GPT-4o, Gemini 2.5 Pro, Claude Sonnet), 8 Visual Reasoning Types (VR01-VR08), pipeline 4 estagios
  • Publicacao open-source: LICENSE (MIT), README.md, CONTRIBUTING.md, .gitignore, GitHub Actions CI (lint+test+LaTeX+artifacts)
  • Validacao clinica em arteterapia decolonial (skill clinical-art-therapy): pipeline 8 estagios, framework 6 dimensoes, protocolo etico CEP/TCLE/LGPD, TCC integrado (Nadielle Darc)
  • Cora-Debate V7: 7 sub-verificadores formais para codigo-fonte (V7a Syntax AST, V7b Logic Prover/Hoare, V7c Type Safety, V7d Resource Bounds/Big-O, V7e Security Patterns/OWASP, V7f Test Coverage, V7g Invariant Checker)
  • Score composto V7 0-100 com pesos por sub-verificador
  • TDD Academic v2.0 validado: 25/25 testes (100%), relatorios JSON+MD
  • Anteprojeto PPGTE validado: PDF compilado (9 pgs, 161KB), apendice TDD inserido
  • Dissertacao compilada: 30 pgs, 381KB, PDF via MiKTeX

Changed

  • Taxonomia: 212 -> 350 tipos (+65%), 27 -> 35 categorias (+30%)
  • Cora-Debate: v1.0.0 -> v1.1.0, V1-V6 -> V1-V7 (7 verificadores, 21 sub-verificadores)
  • SKILL.md Cora-Debate: 211 -> 306 linhas (+95 linhas)
  • Documentacao: CHANGELOG.md, README.md, CONTRIBUTING.md para publicacao aberta
  • CI/CD: GitHub Actions adicionado ao workflow de desenvolvimento

Fixed

  • LaTeX: Unicode █ (U+2588) em Gantt chart corrigido com comando \ganttbar
  • LaTeX: \wedge fora de math mode corrigido para $F^{*}$
  • LaTeX: \multirow pacote ausente adicionado
  • KeyError score no editais-br v7.1: setdefault + CACHE_VERSION

[v4.6.1] - 2026-05-27

Added

  • Artigo Qualis A1 completo (19 paginas, 26 referencias com DOIs auditaveis)
  • Fundamentacao teorica rigorosa: UCB1, Platt Scaling, ECE, Cora-Debate
  • Loop autonomo de micro-versionamento (DETECT->ANALYZE->APPLY->VERIFY->BUMP)
  • 55 problemas IMO testados (2001-2020), PCI medio 98.3/100, 100% aprovacao
  • Dataset IMO integrado: github.com/MarceloClaro/IMO_QUESTIONS_SOLUTIONS
  • ASI-Evolve (GAIR-NLP) como submodule com Cognition Store (10 itens)
  • 212 raciocinios em 27 categorias (8 autonômos: R201-R208)
  • 4 correcoes automaticas: R23 (-56% desativacoes), R34 (-83%), func_eq (+8pp), Platt (ECE 0.25->0.12)
  • Geometry fix: 5->11 reasoning types, PCI 96->98
  • Afiliacao: GeoMaker+IA — Museu Escolar Itinerante (CNM 9.76.35.5698)

Changed

  • Taxonomia: 204->212 raciocinios (novas categorias XXVI Geometric, XXVII Perturbation)
  • Classificador: keyword-based (38%) -> TF-IDF + cosine (70-95%)
  • Calibracao: Platt Scaling integrado ao pipeline de producao
  • Documentacao: 6 novos documentos tecnicos minuciosos
  • README: badges atualizados (IMO, PCI, Cora, Artigo Qualis A1)

[v4.6.1.0] - 2026-05-27

Added

  • Refined the ecosystem structure and improved repository organization.
  • Expanded technical and conceptual documentation for the multi-agent platform.
  • Strengthened the project foundation for future integrations, automations, and operational growth.

Changed

  • Improved the clarity of the ecosystem architecture presentation.
  • Increased project maturity for release management and future versioning.

[v4.0.0] - 2026-05-27

Added

  • Consolidated the multi-agent architecture vision of the OpenCode Ecosystem.
  • Expanded the structured technical artifacts and project documentation.
  • Prepared the repository for scalability across specialized agents and autonomous flows.

Changed

  • Improved overall repository structure and architectural clarity.
  • Strengthened the conceptual and scientific narrative of the platform.

[v1.0.0] - 2026-05-27

Added

  • Initial public release of the OpenCode Ecosystem.
  • Introduced the foundational structure of the platform.
  • Published the first documented vision for a coordinated multi-agent AI ecosystem.

Changed

  • Established the initial base for future evolution of the project.