Skip to content

Releases

v0.1.1

v0.1.1 Latest
Latest

Choose a tag to compare

View all tags
@adeoluo adeoluo released this 09 Nov 01:55
v0.1.1
3cb0570

[0.1.1] 2026-11-09

Bug fixes

  • Current Claude model and Messages API version, with a request shape test (#10)

Documentation

  • Standard README structure and architecture diagram
  • Vertical layered architecture diagram
  • Horizontal layered architecture diagram
  • Compact horizontal architecture diagram
  • Remove authorship statements from docs and scripts
  • Recorded demos of more commands, with their vhs tapes (#3)
  • Recorded demos of the HTTP service and more commands (#4)
  • Complete citation metadata and bump it on release (#6)
  • Render every architecture diagram in the house style (#8)
  • Diagrams on white in the shared theme, make diagram for every diagram, and a self-contained Definition of Done (#9)

Build and CI

  • Run CodeQL on main and on a weekly schedule only
  • Build and attach the sdist and wheel to every release (#11)

Maintenance

  • Add release automation, AGENTS.md, llms.txt and contributor docs
  • Classify squash merges in the changelog by the commits they contain
  • Latest toolchain and CI versions (#5)
  • Release script keeps hard-coded version strings in step (#7)
Assets 5
Loading

agent-memory v0.1.0

agent-memory v0.1.0

Choose a tag to compare

View all tags
@adeoluo adeoluo released this 02 Nov 03:54
v0.1.0
55d62cb

Changelog

v0.1.0 (2026-11-01)

First release.

  • Episodic, semantic, and procedural memory in SQLite with FTS5 and a vector table, isolated by
    namespace.
  • Write path: deterministic rule extractor, pluggable LLM extractor (Claude provider and an
    offline fake), fact deduplication and versioning (reinforce, add, supersede, contest).
  • Read path: hybrid score over BM25, hashing-embedding similarity, recency, and importance, with a
    per-result explanation.
  • Consolidation: extractive session summaries and exponential-decay forgetting.
  • Interfaces: Python API, agent-memory command line (add, recall, consolidate, stats,
    facts, forget, export, import, benchmark), and an MCP server (agent-memory mcp) with
    remember, recall, forget, and list_facts.
  • Benchmark harness with a deterministic multi-session dataset, four memory conditions, a
    scoring-term ablation, development-seed weight tuning, and scaling measurements.
Loading