Repo rescue agent

Copy this into your AI, start the agent, keep the sKache. ๐Ÿ—ƒ๏ธ

The agent runs on the machine where the files live, scans on a ten-minute loop, streams encrypted source-custody artifacts, writes proof receipts, and sends Resend update emails with recovery links.

Open repo lane ๐Ÿ’ป

No perfect-tech promises

ShYT may crash, but Skye keeps the sKache.

No perfect-tech promises. Just proof receipts, recovery prompts, checkpoints, and rollback-aware workflows built to keep your last clean move from disappearing.

ShYT happens, and noo tech is perfect, so despite our extreme stress test and live testing with the owner's own codebase, we will not make the guarantee. We show the receipts, the proof surfaces, and the recovery workflow.

Use this prompt

Copy/paste this into your coding AI ๐Ÿง 

You are setting up SkyeVault Repo Rescue for this workspace.
Read the repo docs and the public SkyeVault agent install page first.
Use the existing FS27/SkyGate/Free99 owner auth lane. Do not create app-specific admin passwords.
Do not print or commit secrets, bearer tokens, passphrases, peppers, signed URLs, env values, or unlock codes.

Goal:
Run a local SkyeVault autosync agent that scans this repo about every ten minutes, tracks Git/dirty/untracked/local-only source state, streams an encrypted source-custody artifact to SkyeVault/R2 when the digest changes, writes public-safe proof receipts, and sends owner Resend emails with proof/vault/unlock links.

Commands to run:
1. npm install
2. npm run vault:agent:status
3. npm run vault:autosync:status -- --env-file=env.txt --mode=full
4. npm run vault:autosync:dry-run -- --env-file=env.txt --mode=full
5. npm run vault:autosync:notify:on
6. npm run vault:agent:start -- --env-file=env.txt --mode=full --interval-seconds=600
7. npm run vault:agent:status
8. npm run vault:autosync:proof

After setup, report:
- whether the daemon/agent is running
- latest autosync receipt path
- latest artifact receipt id if present
- proof page path
- notification status
- any missing env names only, never secret values

Important:
The cloud Worker/R2/vault is the control plane. The local agent is required because only the machine or Codespace can see untracked files, ignored files, local env files, and secret-like source state before it disappears.

Install commands

Local agent setup โš™๏ธ

npm install npm run vault:agent:status npm run vault:autosync:dry-run -- --env-file=env.txt --mode=full npm run vault:autosync:notify:on npm run vault:agent:start -- --env-file=env.txt --mode=full --interval-seconds=600 npm run vault:agent:status npm run vault:autosync:proof
Proof ๐Ÿงพ Autosync proof heartbeat

Public-safe digest, artifact receipt, control-pack receipt, daemon state, and proof log.

Vault ๐Ÿ—ƒ๏ธ Vault drive recovery

Owner-gated vault drive where encrypted repo artifacts and control packs can be recovered.

Unlock ๐Ÿ”“ SkyeSecure unlock surface

Use the control pack, passphrase, and pepper out-of-band to unlock custody material.

Command ๐ŸŽ›๏ธ SkyeVault Command Center

Owner dashboard for receipts, notification controls, vault links, proof, and recovery posture.

How it stays real

Cloud control plane plus local file agent

  • ๐ŸŒฉ๏ธ Worker/R2 stores encrypted artifacts, receipts, proof logs, and recovery links.
  • ๐Ÿ’ป Local agent sees the actual repo files, including untracked source and local-only critical files.
  • ๐Ÿ” Source custody encrypts before upload; public proof never publishes secrets or file bodies.
  • ๐Ÿ“ฌ Resend emails include logo, proof, vault drive, unlock surface, command center, artifact receipt, and control-pack receipt links.

Recovery window

What "about ten minutes" means

If the machine or Codespace stays alive, the agent keeps scanning even when the editor closes. If the whole workspace is destroyed, recovery starts from the latest completed encrypted artifact. Full streams can take longer than ten minutes; a fast encrypted delta journal is the next layer for stricter last-edit capture.