Lifeboat
Give your AI a memory that survives the session, and a backup that protects it all.
AI assistants forget everything when the chat ends, and their files can vanish with one wrong click. Lifeboat is a build you install once: your AI maintains its own memory in plain files, backs them up on a schedule, and runs a guard that makes sure nothing slips through the cracks. How it works: 1. Paste the keeper guide into your AI's instructions. 2. It sets up a small set of files: a curated memory (durable facts, preferences, decisions), an operating manual (how you two work together, lessons learned), and dated journals (what was done, when, with what result). 3. After each session, it files away what matters: a preference you stated, a decision you made, a lesson from something that went wrong. Transcripts and routine chatter stay out; only the durable stuff goes in. And at any time you can say "forget X": it removes that fact from the memory files and confirms. Your memory obeys you, not the other way around. 4. Next time you talk, it reads the files first. No more repeating yourself, no more re-learning the same lessons. 5. It sets up a regular backup of all these files: a local copy plus a private repo, with rotation. A deleted file is never a lost file. 6. It installs a coverage guard: on a schedule, it checks that every important folder is actually in the backup, and reports anything new that isn't covered yet. Started a new project folder? It gets flagged before it's ever at risk. This guard exists because new folders have a habit of silently falling outside backups. 7. Everything is plain text you can read, edit, and back up yourself. Nothing is uploaded anywhere except your own private repo, nothing depends on a vendor's memory feature. 8. Not for you? A clean uninstall prompt is included: your AI lists everything it installed, asks what to keep (your memory files and backups are never deleted without your explicit word), removes the rest, and confirms. What is inside: - SKILL.md: the keeper master prompt (what to remember, what to forget, where everything goes, the end-of-session filing habit, the backup routine). - templates/: blank templates for the memory file, the operating manual, and the journals. - scripts/backup.py: the scheduled backup script (local copy + private repo, rotation included, retries automatically with backoff on transient API errors, reports failures instead of failing silently, secrets stay out of the file list by default). - scripts/coverage-check.py: the anti-gap guard (compares your real folders against the backup list, reports anything new). - examples/: a small filled example so you can see the shape of it before you start. No dependencies beyond Python for the scripts. Works with any agent that accepts custom instructions, and it mirrors your language. Your data stays yours: the templates contain zero personal information, the backup goes to a repo you own, and the guide reminds your AI to keep secrets, credentials, and private details out of the shared files.
Version: 1.0.0
Quick answer
Lifeboat is a build you install once so your AI stops forgetting: it maintains its own memory in plain files — a curated memory of durable facts and decisions, an operating manual for how you work together, and dated journals — files away what matters after each session, and reads the files first next time. On top of that, it sets up scheduled backups to a local copy and a private repo you own, plus a coverage guard that flags important folders the backup does not cover yet.
At a glance
| What it is | A keeper system for your AI: persistent plain-file memory, an end-of-session filing habit, scheduled backups with rotation, and a backup coverage check. |
|---|---|
| Who it's for | Anyone who works with an AI assistant over time and is tired of repeating themselves — or of losing the files that hold the context. |
| What you give it | One paste: the keeper guide goes into your AI's instructions. From there it creates the memory files from blank templates, and backups go to a private repo you own. |
| What you get | An assistant that reads its memory before you start, files decisions, preferences and lessons after each session, removes a fact when you say forget it, and keeps everything backed up — all in plain text you can read, edit and back up yourself. |
| Works with | Any agent that accepts custom instructions, and it mirrors your language. The scripts need only Python; nothing depends on a vendor's memory feature, and nothing is uploaded anywhere except your own private repo. |
| Cost | Free — this listing is free. |
| Time | One installation; after that it runs as a habit — filing at the end of sessions, backups on their schedule, and the coverage guard checking periodically for uncovered folders. |
Example — a decision that survives the weekend
An example of how you might use it, following the workflow the build documents:
- On Friday you settle a question with your AI, and at the end of the session it files the decision — and the lesson behind it — in the memory files.
- The scheduled backup copies those files locally and to your private repo, and the coverage guard confirms your important folders are covered.
- On Monday, a fresh session reads the files first — the decision is simply there, with no recap needed.
Install
Copy the install package below, then paste it into Muse🧪 Install ran in our test sandbox · Nov 11, 2026📌 Pinned version: commit a2f96d9Pinned version: install the code at commit a2f96d90e1c5fc99ca99259fb0eacc401a92fae1 exactly (Skill Harbor tested this version). Do not substitute a newer version. If this commit is unreachable, stop and tell me. Turn yourself into my workflow keeper. Read the keeper guide at https://raw.githubusercontent.com/skill-harbor-hq/lifeboat/a2f96d90e1c5fc99ca99259fb0eacc401a92fae1/SKILL.md and follow it. Set up my AI memory and backup system: a curated memory file, an operating manual for how we work together, dated journals, a scheduled backup of all of it to my own private repo, and a coverage guard that reports any important folder not yet in the backup. Show me the files you create, explain what goes where, and after each session file away what matters without me having to ask. Never store credentials, secrets, or private details I have not explicitly approved for the memory.
Saved to your recent installs. Find it anytime on /connect.
🔒 Security, provenance and what this touches
- It can create or modify files.
What the scan found — excerpts from the install prompt
- writes files — Set up my AI memory and backup system: a curated memory file, an operating manual for how we work together, da
The install prompt below already includes the vetting steps: your agent follows the community checklist before installing anything with executable code. Want more?
Verification certificate
Dated facts about what was checked for this listing, and how. This is not a safety certification.
- Pinned version
- a2f96d9 · 2026-10-13
- Sandbox install test
- Install ran in our test sandbox · Nov 11, 2026
- Creator identity
- Verified · 2026-10-04
- Fingerprint watch
- Last checked · Nov 19, 2026 : no change detected in the files linked by this install prompt
- Install prompt analysis
- 2026-10-13 : writes files
Skill Harbor never reviews the code of a listed skill. These are dated facts about the install prompt and our install test, not a safety certification. Review a skill yourself before you install it. How we verify
Questions
How do I install a skill?
Every product page includes a copy-paste install prompt. Paste it into your Muse and it sets the skill up for you — no manual configuration.
Where does my money go?
Straight to the seller. Skill Harbor never processes payments: checkout happens on the seller’s own page, usually Stripe.
What does the ✓ next to a creator’s name mean?
It means we confirmed the identity of the person behind the listing. It says nothing about the code itself — always check a skill before installing it.