Golden Thread — AI skill for Claude Code
Vault-based AI memory for Claude Code, with hook-enforced Core rules.
How to install Golden Thread
This entry records only its repository, not the path inside it, so there is no
exact command to give. Open shaven/golden-thread and copy the folder into
~/.claude/skills/, or the file into ~/.claude/agents/.
What Golden Thread does
Vault-based AI memory for Claude Code, with hook-enforced Core rules.
Alternatives in AI
- Pro Workflow — 1,400+ Battle-tested Claude Code workflows from power users 1.4k ★
- CLI WeChat Bridge — 将 AI 命令行工具以最为原生的方式,集成到微信 ClawBot / 企业微信 bot 中,目前支持集成 Codex、Claude Code、OpenCode、Pi Agent 478 ★
- Claudexor — Multi-harness control plane for Claude Code, Codex, Cursor, and OpenCode: quota-aware rotation across multiple 425 ★
README
Golden Thread
A memory system for AI coding sessions, built on plain markdown and git — and, unusually, one where the rules that matter most are **mechanically enforced** rather than merely written down.
Every AI session starts with amnesia. You explain your conventions, the session does good work, the session ends, and it is all gone. Golden Thread is one Obsidian vault that acts as shared memory between you and every session you run: sessions read from it at startup, look things up while working, and write back what they learn.
Its distinguishing idea is the second problem, the one most memory systems never address: **writing a rule down does not mean it gets followed.**
Plugin **v0.9.11**. Seven Core rules currently enforced, three of them *validated* — a `Stop` hook inspects the finished reply and blocks it if the rule was broken.
Who this is for
You run several projects at once and the ideas never arrive in the right one. You are deep in project A when the fix for project B occurs to you, a session ends before the finding is written down, and by Monday you cannot remember which of the six things you touched last week was the urgent one. If your attention scatters, from ADHD or simply from load, a system that depends on remembering to file things in the right place will not hold.
Golden Thread is built so that **capture never waits for the right context.** Write the thought as one line in the vault's inbox, from whatever session or project you happen to be in. A review sweep later routes each line to the project it belongs to. Every project keeps its open tasks in one place, and a single generated rollup ranks them across *all* projects, escalating by age and deadline, so "what should I work on?" has one answer instead of six status pages.
The result is one place to look. Sessions read from it at startup, write back what they learned at the end, and the rules you most need to hold are re-asserted every turn instead of trusted to memory.
Feedback
Related Skills
Petro Agent
Autonomous petrophysical reports from raw LAS well logs - the LLM decides, a golden-tested deterministic engin
Sekha
Zero-dependency AI memory system with hook-level rules enforcement for Claude Code
Session Instructions Claude Hook
AI Orchestration: project memory and rules
Fable Director
Token governance for Claude Code: the top model directs, execution goes to the cheapest adequate means. Routin
Aloop
Claude Code plugin for agentic agile + autonomous TDD: hook-enforced AI coding agents that plan sprints, then
Dryas Workflow
Operating model for AI-assisted software delivery: enforced rules, one owner per job, a logged model ladder. R
Related Agents
Part 11: Auto-Capture Hook (Stop Manually Writing Memory)
By Terp - Terp AI Labs --- The vault system from Part 9 is powerful, but there's a gap: you have to manually w
Reliability Engineer
Python delivery agent for quality gates and test infrastructure. Owns test suite (>=80% coverage enforced), go
Bizreq Analyst
Business-requirements agent — authors business-rules.md (a BR-NN decision table) and golden-scenarios.md (SCEN