Samemind — Documentation skill for Claude Code
Keep your decisions, projects, and context when you switch between Claude Code, Cursor, opencode, and the rest — one plain-markdown bundle in your repo.
How to install Samemind
This entry records only its repository, not the path inside it, so there is no
exact command to give. Open alexgrebeshok-coder/samemind and copy the folder into
~/.claude/skills/, or the file into ~/.claude/agents/.
What Samemind does
Keep your decisions, projects, and context when you switch between Claude Code, Cursor, opencode, and the rest — one plain-markdown bundle in your repo. Golden-40 recall@5 0.925 on the live bundle; `samemind install` wires 12 engines; weekly `review` surfaces stale facts before they drift.
Alternatives in Documentation
- 会话管理(Session Manager)需求文档(PRD / Markdown) — 目标:对 Codex / Claude Code 的本地会话记录进行可视化管理,并提供“一键复制 / 一键终端恢复”能力 31.7k ★
- Repo Recap — Generate a structured recap of the repository state: open PRs, open issues, recent releases, and executive sum 11.9k ★
- Ars Format Convert — ARS academic-paper format-convert mode — convert to LaTeX / DOCX / PDF / Markdown 4.5k ★
README
samemind
**Use Claude Code in the morning, Cursor by day, opencode at night — and every agent remembers your decisions, your projects, your context.** It's just markdown in your git: read it with your eyes, edit it by hand, own it forever. Golden-40 recall@5 **0.925** on the live bundle; `samemind install` wires **12 engines**; weekly **`review`** keeps stale facts from living as equals of their corrections. Wire-compatible with [Google OKF v0.1](docs/interop.md).
[](https://github.com/alexgrebeshok-coder/samemind/actions/workflows/ci.yml)
Why not "just markdown + BM25"?
| Typical git-markdown memory | samemind | |
|---|---|---|
| Wire format | ad hoc | OKF v0.1 export/import |
| Identity | flat notes | Identity / User / EngineRule → budgeted brief |
| Work | separate tools | event ledger + board in the same bundle |
| Multi-engine ops | assumed, never checked | fleet registry + heartbeat: who reports, who went quiet |
| Engines | often one client | samemind install → 12 engines (adapters) |
| Freshness | manual copy-paste into context | opt-in daemon + lifecycle hooks, honest per-engine tier |
| Face | raw files | samemind ui — local dashboard: board, graph, fleet, live feed |
Proof (measured, not promised)
**Golden-40 recall bench** — 40 natural-language paraphrase questions (no title tokens) with hand-labeled golden ids, run against the live samemind development bundle (212 indexed docs, bge-m3 embeddings):
| mode | recall@5 | precision |
|---|---|---|
| bm25 | 0.925 | 0.900 |
| semantic | 0.925 | 0.900 |
| hybrid | 0.925 | 0.875 |
Independently verified at release (separate engine, branch replay): hit@3 0.90–0.925 across all three modes; test suite 1386/1386 green.
The road to 0.925: in 1.1.0 the hybrid mode scored **0.275** on the sam
Related Skills
Agent Memory Kit
One memory for every coding agent — plain markdown in your repo. Injected on Claude Code & Cursor, skills on C
MarkSight
Browser-only Markdown editor that packages any document into a Claude Agent Skill (SKILL.md / .skill bundle) i
Tuiboard
Terminal kanban on plain markdown, with optional Today/Tomorrow planner, 24h agenda + calendar, and a live Cla
Observational Memory
Local-first shared memory for Claude Code, Codex, OpenCode, Kimi, Grok, Cowork, and Hermes, with bounded obser
Stele
Rate limited mid-task? Crashed mid-edit? Locked into one harness by the context it holds? A skill that keeps t
MushroomTW Skills
Five portable Agent Skills for README authoring, dependency decisions, repository bug audits, and SonarQube qu
Related Agents
Ff l10n
Flapforge localization engineer: adds and audits every player-facing string in en and pt_BR, keeps the two fil
Text Slop Cleaner
This agent should be invoked to remove machine sounding text and unnecessary comments. This includes rewriting
Scope Ingest
Bounded subagent that turns raw source material (markdown files, plain-text files, pasted text) into a normali