Claude Memory Kit

Claude Memory Kit v4

Persistent memory for Claude Code agents. Layered memory, agent-driven promotion, zero manual file editing.

Two core invariants

1. User only talks. Agent captures, proposes, writes. Every architectural decision passes this test.
2. Every memory entry carries a [YYYY-MM-DD] date tag. This is what lets /close-day detect cross-session repetition and propose promotions.

What's different from v3.2

• Agent-driven promotion ritual via /close-day (was: background promote-patterns.py detection + flush.py auto-flush, both killed as unreliable and invariant-violating)
• Multi-project isolation via projects// — shared layers load always, per-project scope on demand
• experiments/-YYYYMMDD/ sandbox layer next to projects/ — different lifecycle, no direct promotion to rules
• Date-tagging promoted to a documented load-bearing invariant (was implicit machinery, now stated)
• Killed: experiences/ staging layer, promote-patterns.py + flush.py background scripts, playbooks/ and -guidance/ reference-skill seeds, the /memory-audit and /memory-compile operators, and the knowledge/connections/ + knowledge/meetings/ subdirs. The role-guidance pattern still works if you add it per-project (see .kit/ARCHITECTURE.md § "Adding role-guidance yourself"); the kit just doesn't ship templates.
• Default surface trimmed to two operators (/close-day + /tour). The wiki-maintenance commands (/memory-lint, /memory-query) plus a usage-telemetry tool (/memory-usage) moved to opt-in .kit/advanced/ — see its README to enable.

See [.kit/CHANGELOG.md](.kit/CHANGELOG.md) for full migration notes.

Quick start

git clone https://github.com/awrshift/claude-memory-kit.git my-project
cd my-project
claude

First session: agent greets you, asks 2-3 setup questions, loads you in. Type /tour for a guided walkthrough.

Daily workflow

1. Open a session — hooks auto-load NSP + MEMORY + knowledge index
2. Work normally — agent captures patterns in MEMORY.md as you speak
3. /close-day when done — agent synthesizes today, audits for promotions, proposes verbally, writes on your verbal "yes"

Tomorrow starts where today left off.

Included skills

Default — the whole daily loop:

| Skill | Description | |---|---| | /close-day | End-of-day audit ritual: synthesis + promotion proposals. Auto-backfills missed working days from git history. | | /tour | Interactive guided walkthrough on your own files |

Opt-in (.kit/advanced/, copy into .claude/ to enable):

| Skill | Description | |---|---| | /memory-usage | Read-only telemetry: hot files vs cold archival candidates | | /memory-lint | Structural hygiene (broken links, sparse articles, orphans) | | /memory-query | Natural-language search across the knowledge base |

Architecture

Memory lives in layers, each answering a different question:

| Layer | Answers | Written by | |---|---|---| | daily/YYYY-MM-DD.md | «what happened today» | agent via /close-day | | .claude/memory/MEMORY.md | «what patterns repeat across sessions» | agent as you speak | | knowledge/concepts/*.md | «facts and rationale by topic» | agent after your "yes" on /close-day | | .claude/rules/*.md | «what must always / never happen» | agent after a long-stable pattern |

Promotion runs liquid → amber → crystal: an observation in daily/ becomes a date-tagged pattern in MEMORY.md, which /close-day may promote to a knowledge/concepts/ article or a .claude/rules/ constraint — always agent-written, always on your verbal confirmation.

See [.kit/ARCHITECTURE.md](.kit/ARCHITECTURE.md) for full details and [CLAUDE.md](CLAUDE.md) for the agent's session workflow.

Built from production use

Iteration on 700+ real sessions across 7+ projects. Every component earns its place; experiences/, background-detection scripts, and generic role-guidance seeds didn't survive review.

Source & license

This open-source skill is cataloged on AgentStack and links to its original source — we do not rehost the code.

• Author: awrshift
• Source: awrshift/claude-memory-kit
• License: MIT

Install and usage instructions live in the source repository linked above.