Changelog¶

All notable changes to aptdata are documented here.

The format follows Keep a Changelog and this project adheres to Semantic Versioning.

[Unreleased]¶

Multi-agent core (aptdata.agents) — IAgent/AgentRegistry, Router (prefix/skill/llm/default), Project/ProjectRunner, adapters reais (OpenClaw, ClaudeCode, OpenCode via CLI) e MCP dispatch. Novos comandos aptdata agents (list/send/route/dispatch/resolve) e aptdata project (init/plan/run).
Google Calendar OAuth2 — New aptdata.integrations module with OAuth2 client for Google Calendar API.
Analytics Engine — Migrated analytics and gamification engine from mindflow (aptdata.analytics.indicators, stats, clusters, dedup, TF-IDF).
Gamificação — XP, streak e levels portados do mindflow para aptdata.gamification.
Guardrails — .gitignore for secrets and .env.example template.

Observabilidade — Added timed_create to __init__.py exports.
CI — Lint errors and test dependency failures resolved across multiple PRs.
Lint — Zerada a dívida de ruff em aptdata/ e tests/ (imports, linhas longas, __all__ nos re-exports).

Quick start example — Updated to reflect current API with IExecutor/DefaultExecutor.

docs/prompts/docs-agent.md — Authored the DevPortal Docs Agent prompt, establishing strict guidelines for Brazilian Portuguese language, anti-prolixity, MkDocs Material features, and documentation architecture.
IExecutor and DefaultExecutor — Added graph execution decoupling from BaseWorkflow.
IAsyncComponent — Added native async/await support for I/O bound pipelines and LLMs.
IIterableDataset — Added generator-based datasets to support lazy evaluation and streaming contracts.

DevPortal Rewrite — Overhauled all core documentation (index, getting-started, architecture, scaffold-templates, configuration, telemetry, quality, governance, transform-engines, diys, ci-logs, mcp) translating content to Brazilian Portuguese.
UI/UX Evolution — Implemented MkDocs Material admonitions, content tabs, and grids across documentation files to improve Developer Experience (DX).
Navigation Audit — Re-structured the mkdocs.yml navigation tree into Getting Started -> Core -> Plugins -> API. Enabled the md_in_html extension for layout support.
IDataset Generics — Refactored IDataset to inherit from Generic[T], enforcing strict typing contracts.
StateBackend Optimization — Removed raw record materialization from workflow checkpoints, persisting only metadata to prevent OOM errors in Big Data flows.

aptdata.core.workflow — context-aware workflow module introducing:
IWorkflow / BaseWorkflow with before_run / after_run hooks.
WorkflowNode / WorkflowEdge graph primitives with optional conditional routing.
compile() adjacency-list precomputation and cycle validation.
run() execution with hook lifecycle and ExecutionContext integration.
tests/test_workflow.py — tests covering workflow primitives, lifecycle, compile validation, and conditional execution behavior.

system.py — universal architecture module introducing:
ComponentKind enum (TRANSFORM, FILTER, AGGREGATE, EXTRACT, LOAD, CUSTOM).
ComponentMeta dataclass — rich metadata (kind, tags, branch_on, description, extra).
IComponent / BaseComponent — replaces IStep/BaseStep; components return list[IDataset] enabling multi-output and branching flows.
FlowEdge — directed edge with optional conditional predicate.
FlowNode — graph node with back-reference to its owning IFlow.
IFlow / BaseFlow — directed execution graph with add_component(), connect(), compile() and run().
ISystem / BaseSystem — top-level orchestrator with register_flow() and run().
tests/test_system.py — 34 tests covering all new abstractions.

aptdata.plugins: _PipelineRegistry renamed to _SystemRegistry; list_pipelines() renamed to list_systems().
aptdata.cli.app: CLI now instantiates systems with system_id=<name>.
aptdata.core.__init__: exports updated to the new architecture.
tests/test_core.py: stripped to dataset-only tests.
tests/test_cli.py: mock updated from _MockPipeline to _MockSystem.
All documentation updated to reflect the new System/Component/Flow model.

Two-layer contract system – pure @dataclass + ABC interfaces (IDataset, IStep, IPipeline) and Pydantic-validated base classes (BaseDataset, BaseStep, BasePipeline).
Plugin registry – aptdata.plugins.registry singleton for registering and discovering concrete pipeline implementations by name.
CLI – aptdata run command: runs a registered pipeline and emits structured JSON events to stdout/stderr.
CLI – aptdata monitor command: launches the interactive Textual TUI monitoring dashboard.
TUI – MonitorApp with DAG panel, per-step status table and memory-usage bar; auto-refreshes at a configurable interval.
Test suite – pytest-based tests covering all core interfaces, base classes, the CLI and the plugin registry.
Initial project documentation.