More screenshots
Works with the AI Agents you already use
brain
Connect Agent instances via a private GitHub repo to maintain persistent project memory (Optional visual mind map when running with claude).
$6.99
brain
Example session with this skill installed
brain read – where did we leave off with the kitchen renovation? I worked on it on my laptop yesterday and want to continue on my phone.
brain save – we decided to go with the oak countertop, the electrician comes on March 3, and the tile order is done.
brain map – show me everything in my memory as a mind map, including what I've already finished.
- Read your context and instructions
- Compiled the brain
Memory loaded – overview.md and projects/kitchen-renovation.md (last updated yesterday, 21:14, from Claude Code on your laptop). Here's where you left off with the kitchen renovation:
Summary
Planning is about 70% done. The layout and countertop are decided, the electrician is booked, and the tiles are the one big open decision. Nothing is ordered yet except the sink.
Current status
| Area | Status | Details |
|---|---|---|
| Layout | ✅ Decided | L-shape, island dropped (too tight for the walkway) |
| Countertop | ✅ Decided | Solid oak, 40 mm, oiled – chosen over quartz for price |
| Sink | ✅ Ordered | Undermount, single bowl, delivery week of Feb 24 |
| Electrician | ✅ Booked | March 3, 8 am – 4 new sockets + under-cabinet lights |
| Tiles | ⏳ Open | Two samples left: white zellige vs. sage green subway |
| Cabinets | ⏳ Open | Waiting for final quote from the carpenter |
| Budget | ⚠️ Watch | €11,400 planned, €8,950 committed so far |
Decisions you made yesterday
- Dropped the island so the dishwasher door can open fully
- Oak instead of quartz saved about €1,100
- Under-cabinet lighting gets its own switch, not tied to the ceiling light
Still open
- Pick the tiles – you wanted to see both samples in daylight first
- Chase the carpenter's quote (asked last Tuesday)
- Decide whether the old fridge stays or gets replaced
Next steps
- Since you're on your phone: take a daylight photo of both tile samples and send them here, and I'll help you compare them
- Once the cabinet quote arrives, say "brain save" with the price so the budget stays current
- The electrician needs the final socket positions before March 3 – want me to draft that list from your layout notes?
Connects securely to your tools. The creator never sees your data.
About this skill
Claude forgets your project context the moment you close a chat. You lose decisions, progress updates, and project history across different devices. This skill establishes a persistent, shared memory layer using a private GitHub repository as the backend.
What it does
- Multi-surface sync allows the Claude mobile app, web interface, and Claude Code to share the same project state.
- State persistence saves decisions, project status, and task lists into Markdown files in a private repo.
- Automated updates capture new insights and milestones from conversations and commit them to version control.
- Context loading pulls the latest project overview and relevant files at the start of a session.
- Visual mapping generates an optional mind map artifact to browse the entire memory structure.
How it works
- Run
brain configureto link a GitHub account, create a private repository, and initialize the Markdown structure. - Use
brain readat the start of any conversation to pull the current status of your projects. - Call
brain saveafter making decisions or finishing tasks to sync changes back to GitHub. - Use
brain mapto visualize project connections and status through an interactive mind map.
Frameworks & tools
The system uses GitHub for storage and version control. It integrates with the Claude GitHub App and Claude Artifacts. Local environments require Git and Claude Code.
Why this beats prompting it yourself
Manual context management leads to stale data and fragmented notes across chats. This skill automates the Git workflow, handles merge conflicts, and ensures every Claude instance you use is aligned with the same source of truth.
Use cases
- Tracking long-term project decisions across different coding sessions.
- Syncing ideas captured on the mobile app with your desktop dev environment.
- Maintaining a high-level overview of multiple active workstreams.
- Preserving a "Done" history that survives chat window closures.
Known limitations
Requires a GitHub account and installation of the Claude GitHub app. The mind map feature requires an environment that supports private artifacts with database capabilities.
Find more skills like this on the Claude Skills Marketplace, the Codex Skills Marketplace and the Gemini Skills Marketplace, or browse Claude Skills for Data & Databases.
Requires a free GitHub account and a private repo; the memory lives there, not on Agensi.
- Your agent needs git or a GitHub integration with write access. Without write access, "brain read" works but "brain save" cannot store anything.
- The memory map (brain map) works only in Claude with artifacts, e.g. Claude Code in the Claude desktop app. Other agents skip it; the next save from Claude catches up.
- Memory is not automatic: you say "brain read" to load it and "brain save" to store new decisions. Nothing is saved in the background.
- Setup takes about 10–15 minutes the first time (GitHub repo, access, one pointer line per agent). Non-developers are guided step by step.
- Each agent you use needs the pointer line in its own instructions once, otherwise it has to ask where the memory is.
- Designed for one person's memory. Several people writing to the same repo can work, but there are no roles or permissions beyond GitHub's.
- If two devices save at the same time, the agent has to merge the changes. It is told never to discard anything, but you may occasionally be asked to confirm.
- Large memories are read selectively (overview plus the relevant project files), so very detailed questions about rarely used projects may need a second "brain read".
- Tested end to end with Claude. Codex CLI, Gemini CLI and Cursor follow the same instructions but are less tested.
How to use
- 1
Connect your agent to the Agensi MCP
Copy the prompt and paste it into Claude, ChatGPT, Cursor or any agent that supports MCP. It sets up the connection for you. Manual setup
- 2
Sign in to Agensi
Your agent shows a sign-in link. Sign in and click Allow.
- 3
Ask for what you need
Your agent loads this skill and uses it on your task. It always asks before buying a skill.
Reviews
No reviews yet
Be one of the first to try it. Every listed skill goes through the trust checks below.
Security scanned
Checked by our 8-point scan before listing
Fresh listing
Recently published to Agensi
Money-back guarantee
Doesn't do what its page promises? You get your money back.
Trust & safety
Security scan
Security scan passed today
- Accounts
GitHub
- Software
Git
Permissions
Allowed Hosts