explaingit

hacksurvivor/pathmark

Analysis updated 2026-05-18

2TypeScriptAudience · developerComplexity · 2/5Setup · moderate

TLDR

A shared local memory layer for AI coding tools like Claude Code, Codex, and Cursor, so decisions and project rules saved in one tool are available in all the others.

Mindmap

mindmap
  root((Pathmark))
    What it does
      Shared memory layer
      Cross-tool context
      Local JSONL store
    Tools supported
      Claude Code
      Codex
      Gemini CLI
      Cursor
    MCP Tools
      remember
      recall_memory
      search_memory
      get_context
    Setup
      npm install
      Per-harness config
      No account needed
Click or tap to explore — scroll the page freely

Code map

Detail Auto

An interactive map of this repo's files and how they connect — its source is parsed live in your browser. Click Visualize to build it.

filefunction / class

What do people build with it?

USE CASE 1

Save architectural decisions and project rules once so they are available in Codex, Claude Code, Gemini CLI, and Cursor without re-explaining each session.

USE CASE 2

Switch between AI coding tools during a single session and carry notes automatically from one tool to the next via the shared memory store.

USE CASE 3

Search your saved project memory to recall past decisions when picking up a task after a long break.

USE CASE 4

Migrate an existing Codex memory store to Pathmark using the import script without losing any past entries.

What is it built with?

TypeScriptNode.jsMCPJSONL

How does it compare?

hacksurvivor/pathmarkarashthr/hugo-flowargeneau12e/kairos-tx
Stars222
LanguageTypeScriptTypeScriptTypeScript
Setup difficultymoderatemoderatehard
Complexity2/53/54/5
Audiencedeveloperdeveloperdeveloper

Figures from each repo's GitHub metadata at analysis time.

How do you get it running?

Difficulty · moderate Time to first run · 30min

Requires adding Pathmark as an MCP server separately in each AI coding tool (Claude Code, Codex, Cursor, etc.).

In plain English

When you use AI coding tools like Claude Code, Cursor, Codex, or Gemini CLI, each one starts with no knowledge of what you decided in a previous session or what another tool already knows about your project. Pathmark solves this by giving all of these tools a single shared memory store that lives as a plain text file on your own computer. The memory store is a JSONL file at ~/.pathmark/memory/memory.jsonl (a file where each line is a self-contained record). Each of your AI tools connects to Pathmark through a standard MCP (Model Context Protocol) server. When one tool saves a decision or a project rule using the "remember" command, any other tool connected to the same store can retrieve it in the next session using "search_memory" or "recall_memory". There are also commands for saving higher-confidence conclusions, listing past decisions, and deleting entries you no longer need. Setting it up requires installing Pathmark globally with npm and then adding it as an MCP server in each AI tool you use. The project includes a setup command that generates the correct configuration snippet for Claude Code, Codex, Gemini CLI, Cursor, and several others. For Codex users, there is an auto-capture mode that automatically records context at the start of each session. The tool does not require an account, API key, or cloud service. All data stays on your machine in plain files that you can inspect, back up, or delete. There is also an import script for migrating memory from older Codex-compatible memory stores without overwriting anything. Pathmark is written in TypeScript, distributed as an npm package, and is aimed at developers who regularly use multiple AI coding tools on the same project and want those tools to share context without manual copy-pasting between sessions.

Copy-paste prompts

Prompt 1
Set up Pathmark so Claude Code and Codex share the same memory store. Show me the install commands and the config for each tool.
Prompt 2
How do I use Pathmark's 'remember' and 'recall_memory' tools inside Claude Code to save and retrieve project decisions across sessions?
Prompt 3
I want Pathmark to auto-capture context at the start of every Codex session. How do I install the Codex auto-capture adapter?
Prompt 4
Help me migrate my old Codex memory store to Pathmark using the import script. What flags do I need to avoid overwriting anything?
Prompt 5
How do I configure Pathmark to use a local CLI model for synthesis instead of letting the MCP client's own model answer questions?

Frequently asked questions

What is pathmark?

A shared local memory layer for AI coding tools like Claude Code, Codex, and Cursor, so decisions and project rules saved in one tool are available in all the others.

What language is pathmark written in?

Mainly TypeScript. The stack also includes TypeScript, Node.js, MCP.

How hard is pathmark to set up?

Setup difficulty is rated moderate, with roughly 30min to a first successful run.

Who is pathmark for?

Mainly developer.

Open on GitHub → Explain another repo

This repo across BitVibe Labs

Scan in gitsafehub Deploy in gitdeployhub hacksurvivor on gitmyhub

Verify against the repo before relying on details.