From 0d31563feb2813a45ded27fcfa5a82f2d88d196d Mon Sep 17 00:00:00 2001 From: GAP Promoter Date: Tue, 26 May 2026 03:21:53 +0000 Subject: [PATCH] Add coleam00/claude-memory-compiler to the registry --- .../README.md | 67 +++++++++++++++++++ .../metadata.json | 24 +++++++ 2 files changed, 91 insertions(+) create mode 100644 agents/coleam00__claude-memory-compiler/README.md create mode 100644 agents/coleam00__claude-memory-compiler/metadata.json diff --git a/agents/coleam00__claude-memory-compiler/README.md b/agents/coleam00__claude-memory-compiler/README.md new file mode 100644 index 0000000..d7b05ae --- /dev/null +++ b/agents/coleam00__claude-memory-compiler/README.md @@ -0,0 +1,67 @@ +# claude-memory-compiler + +> Give Claude Code a memory that evolves with your codebase. + +**claude-memory-compiler** is an autonomous Claude Code agent that hooks into +your session lifecycle to silently capture, compile, and inject a growing +personal knowledge base — inspired by [Andrej Karpathy's LLM Knowledge Base +architecture](https://gist.github.com/karpathy/442a6bf555914893e9891c11519de94f), +adapted for AI-assisted coding. + +## What It Does + +Every time you end a Claude Code session (or it auto-compacts), the agent: +1. **Captures** the conversation transcript via the `SessionEnd` / `PreCompact` hooks +2. **Extracts** decisions, lessons, patterns, and gotchas using the Claude Agent SDK (`flush.py`) +3. **Appends** a structured summary to today's daily log in `daily/YYYY-MM-DD.md` +4. **Compiles** daily logs into cross-referenced knowledge articles under `knowledge/` (nightly, or on demand) +5. **Injects** the knowledge-base index + recent log back into the next session via the `SessionStart` hook + +The result: Claude always "remembers" your architecture decisions, preferred patterns, and hard-won lessons — without you having to re-explain them. + +## Key Capabilities + +| Script | What it does | +|---|---| +| `hooks/flush.py` | Extracts key knowledge from a session transcript via Claude Agent SDK | +| `scripts/compile.py` | Compiles daily logs into concept / connection / Q&A articles | +| `scripts/query.py` | Answers questions using index-guided retrieval (no RAG) | +| `scripts/lint.py` | Runs 7 health checks (broken links, orphans, contradictions, staleness) | + +## Why No RAG? + +Karpathy's insight: at personal scale (50–500 articles), an LLM reading a +structured `index.md` outperforms cosine similarity. No vector database, no +embeddings — just clean markdown and a master index. + +## Quick Start + +Tell your AI coding agent: + +``` +"Clone https://github.com/coleam00/claude-memory-compiler into this project. +Set up the Claude Code hooks so my conversations automatically get captured +into daily logs, compiled into a knowledge base, and injected back into future +sessions. Read the AGENTS.md for the full technical reference." +``` + +## Requirements + +- Claude Code with a Max, Team, or Enterprise subscription (no separate API billing) +- Python 3.11+ with `uv` + +## Architecture + +``` +Conversation + → SessionEnd/PreCompact hooks + → flush.py (Claude Agent SDK) extracts knowledge + → daily/YYYY-MM-DD.md + → compile.py → knowledge/concepts/, connections/, qa/ + → SessionStart hook injects index into next session + → cycle repeats +``` + +## Repository + +https://github.com/coleam00/claude-memory-compiler diff --git a/agents/coleam00__claude-memory-compiler/metadata.json b/agents/coleam00__claude-memory-compiler/metadata.json new file mode 100644 index 0000000..aa55192 --- /dev/null +++ b/agents/coleam00__claude-memory-compiler/metadata.json @@ -0,0 +1,24 @@ +{ + "name": "claude-memory-compiler", + "author": "coleam00", + "description": "Turns Claude Code sessions into a self-organising knowledge base — hooks capture transcripts, Claude Agent SDK extracts lessons, compiles cross-referenced articles. No RAG needed.", + "repository": "https://github.com/coleam00/claude-memory-compiler", + "path": "", + "version": "1.0.0", + "category": "developer-tools", + "tags": [ + "memory", + "knowledge-base", + "claude-code", + "productivity", + "hooks", + "agent-sdk", + "knowledge-management", + "personal-ai" + ], + "license": "MIT", + "model": "claude-sonnet-4-5-20250929", + "adapters": ["claude-code", "system-prompt"], + "icon": false, + "banner": false +}