Local, git-versioned memory for AI coding agents. No RAG, no Docker, no external service. Capture, compile, recall over a local LLM wiki with on-device embeddings and an MCP server.
Latest release v0.5.0 — v0.5.0 — startup briefs, glance-view recall, and a Recently reminder · 5 Jul 2026
6 files for Claude Code, Codex and OpenCode: debug-capture-pipeline, run-tests-safely, triage-review-findings, verify-live-change and 2 more — 2,138 tokens loaded in every session.
AGENTS.md A 2,138 tok .claude/skills/debug-capture-pipeline/SKILL.md A 56 tok .claude/skills/run-tests-safely/SKILL.md A 47 tok .claude/skills/triage-review-findings/SKILL.md A 55 tok .claude/skills/verify-live-change/SKILL.md A 60 tok .claude/skills/write-migration/SKILL.md A 60 tok These files are ctxr-dev/llm-wiki-memory's own configuration — they tell Claude Code, Codex and OpenCode how to work on this repository, so they are not mods to install elsewhere. Copy one as a starting point and replace the parts that are about this project.