Skip to content

CHANGELOG.md

Latest commit

 

History

History
63 lines (50 loc) · 3.82 KB

CHANGELOG.md

File metadata and controls

63 lines (50 loc) · 3.82 KB
Raw

Changelog

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

The format is based on Keep a Changelog, and this project adheres to Semantic Versioning.

Unreleased

0.2.3 — 2026-06-10

Added

  • read list --where field=value — filter rows by field value (exact match; repeat for AND). * and ? wildcards supported (e.g. --where name=*Tiexin*).
  • Efficiency benchmarks and regression tests for --where filtering (scripts/benchmark_efficiency.py, tests/test_efficiency.py).
  • New application note: application-notes/llm-sysml-v2-modeling.md — LLM-assisted SysML v2 textual modeling (6U CubeSat PDU controller) following the 6-step pipeline; README and LLM-GUIDE updated with pointers.

0.2.0 — 2026-06-10

Added

  • Single source of truth for the package version: pyproject.toml reads it from src/memnet/__init__.py via hatch.version.
  • memnet version --json for automation; default output is now a wire line @VER: memnet|<version>.
  • memnet add — create new rows only (id_exists if the id is already in the graph).
  • memnet update — replace existing rows only (not_found if the id is missing).
  • CHANGELOG.md.

Changed

  • CLI memnet version now emits the wire-format @VER: memnet|<version> line instead of the plain memnet <version> string.
  • Breaking: memnet write removed. Use add for new rows and update for changes. LLM-GUIDE.md and README.md updated.
  • memnet examples write renamed to memnet examples add.
  • LAW02 in the bundled workflow example now documents add-then-update id discipline.

0.1.0 — 2026-06-10

Initial public release.

Added

  • In-memory working-memory graph for LLM agents (no automatic disk state).
  • memnet serve — local TCP server holding sessions in RAM; CLI is a stateless client.
  • Wire-format I/O (@TAG: field|field|...) on stdout; @ERR, @WRN, @STAT, @DEL signals on stderr.
  • Session lifecycle: session open / resume / current / list / close, TTL (default 60 min, env MEMNET_SESSION_TTL_MINUTES).
  • Optional snapshot save/load (session save --file, session load --file [--ttl] [--keep-id]) — wire-format files the user owns.
  • Tag schema (fixed EDG / LAW + user tags at open), strict field validation, relation allow-list with --allow-new-relation.
  • Graph reads:
    • query warm --anchor <id> — active-only, LAW-prepended (the recommended agent read).
    • query context — cold/full view, warns on stale rows.
    • query neighbors, query path.
  • Direct reads: read list, read get.
  • Mission lifecycle via recycle labels (persistent, delete_on_settle, delete_on_expire); settled missions hidden from warm reads.
  • Housekeeping: housekeep stats|stale|recyclable|dangling|orphans, housekeep prune ... --apply.
  • Advisory warnings: near_cap*, ttl_expiring, stale_in_store, stale_dangling, stale_orphans, stale_graph, mission_settled, fanout_clamped, dangling_endpoint, corrupt_row, etc.
  • Bundled examples (memnet examples map|workflow|write|path|agent-guide) and inline guides (memnet guide, memnet guide --loose).
  • LLM-GUIDE.md — full agent playbook (goldfish loop, settlement pattern, ID discipline, common failure modes).
  • 26 tests passing on Python 3.11 / 3.12.

Notes

  • v1 is local only — TCP loopback, no remote / MCP / authentication.
  • Caps are configurable via MEMNET_MAX_* env vars.
  • Sessions live in process memory only. On serve restart, all sessions are gone unless saved via session save.