Skip to content

v0.2.0-beta

Latest
Latest

Choose a tag to compare

View all tags
@github-actions github-actions released this 28 Sep 13:08
v0.2.0-beta
303e693
This commit was created on GitHub.com and signed with GitHub’s verified signature.
GPG key ID: B5690EEEBB952194
Verified
Learn about vigilant mode.

ContextMemory 0.2.0-beta — self-host launch

Self-hosted memory gateway for your llama.cpp / vLLM server. One OpenAI-compatible /v1 URL. Memory you can open as markdown.

Try it (3 commands)

git clone https://github.com/Kortexio/ContextMemory.git && cd ContextMemory
docker compose -f docker-compose.yml -f docker-compose.llamacpp.yml up --build -d
./scripts/aha-chat.sh

The second chat request carries only the new question — the answer comes from gateway memory. CI runs the same check against a real llama-server on every push.

Highlights

  • Engines: docker-compose.llamacpp.yml (CPU) and docker-compose.vllm.yml (GPU), with tool-calling flags documented (--jinja, vLLM parsers)
  • Chat aha: scripts/aha-chat.sh / .ps1 — two-turn session recall without client-side history
  • e2e CI: e2e-llamacpp badge on main
  • Docs: README leads with self-host; Engines section in docs/self-host.md
  • Hygiene: cleaner Issues (no bot cadence posts), Conventional Commit PR titles, issue templates

Changelog (selected)

Features

  • Bundle llama.cpp and vLLM compose overrides with an end-to-end memory check
  • Self-host launch readiness (README, engines, e2e, public signals)
  • Agentic: raise wiki call budget to 5 distinct queries

Bug fixes

  • Agentic wiki-budget loops and repeated response sections
  • MCP NDJSON framing for azure-monitor-mcp
  • Migrations: cast ConfigJson updates to jsonb
  • ArgumentOutOfRange on agentic string truncation

Full compare: v0.1.0-beta...v0.2.0-beta

Note: This is still a beta. Session memory works with small instruct models (CI uses 1.5B). The full agentic tool loop needs larger models with native tool calling.

Assets 2
Loading