Skip to content

Flow Memory v0.1.0

Choose a tag to compare

View all tags
@Harryanhuang Harryanhuang released this 30 Jun 17:32
v0.1.0
9231127

🎉 Initial Alpha Release

Flow Memory is a SQLite/Postgres/file-backed memory system for AI agents, extracted from the EduFlow Team project.

Features

  • 23 MCP tools — search, profile, sensitive data, packet assembly
  • 3 storage backends — SQLite (default), Postgres, Markdown files
  • LanceDB vector search — with hybrid FTS+Vector fusion
  • V3 features — pin/decay/subject/dual_query/dashboard/AGENTS.md/reflect/skill_evolution
  • Sensitive data encryption — AES-256-GCM with PBKDF2 + security question recovery
  • Configurable policies — MemoryTaxonomy + PromotionPolicy abstractions
  • i18n ready — English + Chinese templates

Installation

```bash
pip install flow-memory
pip install flow-memory[mcp] # + MCP server
pip install flow-memory[vector] # + LanceDB
pip install flow-memory[postgres] # + Postgres
pip install flow-memory[all] # everything
```

Migration from EduFlow

Existing EduFlow users: set `FLOW_MEMORY_DB=~/.eduflow/eduflow_memory.db` and continue using `eduflow.memory.*` imports via the backwards-compat shim. See migration guide.

Test Coverage

  • 19 flow_memory unit tests (storage backends, policies)
  • 514 EduFlow tests pass via the shim
  • Total: ~533 tests passing

See CHANGELOG.md for details.

Assets 2
Loading