GitClaw
Back up the OpenClaw agent workspace to a GitHub repo and keep it synced via a cron-driven commit/push script.
Mariano Pardo
@marian2js
What This Skill Does
Automates backing up an OpenClaw agent workspace to a GitHub repository using a cron-driven commit and push script. Handles prerequisite installation (git, cron, GitHub CLI) and sets up periodic syncs without requiring a personal access token.
Replaces manual Git backups and cron setup by auto-installing dependencies and configuring scheduled pushes to GitHub with minimal user input.
When to Use It
- Set up automatic hourly backups of the agent workspace to a new GitHub repo
- Configure a private GitHub mirror for sensitive agent data
- Enable periodic syncs without needing to remember manual commits
- Restore agent workspace from a previous GitHub backup
- Migrate agent data to a different machine using the synced repo
Install
$ openclaw skills install @marian2js/gitclawGitClaw: Agent self-backup to GitHub
When to use
Use this skill automatically when it's installed or reference, and also when the user asks to:
- back up / sync / mirror the OpenClaw workspace
- set up periodic commits and pushes
- install/configure GitHub backup automation for the agent
Behavioral rule (important)
- Do everything automatically and quietly.
- Only message the user when:
- you need inputs (repo name, visibility, optional frequency), or
- Git/GH/Cron install fails, or
- GitHub auth requires the user to complete the browser/device step, or
- Git operations fail (conflicts, permissions, etc).
Inputs to ask the user for
Ask the user for:
- Repository name (e.g.,
openclaw-backup) - Visibility:
publicorprivate - (Optional) Backup frequency in minutes (default: 60)
Do NOT ask for a Personal Access Token. We use gh auth.
Step 1: Ensure prerequisites (auto-install)
1A) Ensure git is installed (auto-install; do not stop)
Check:
git --version
If missing, install using the best available method:
- If
brewexists:brew install git
- Else if
apt-getexists:sudo apt-get update && sudo apt-get install -y git
- Else if
dnfexists:sudo dnf install -y git
- Else if
yumexists:sudo yum install -y git
- Else if
pacmanexists:sudo pacman -S --noconfirm git
- Else if
zypperexists:sudo zypper install -y git
- Else if
apkexists:sudo apk add git
- Else if macOS and
xcode-selectexists:xcode-select --install(this may prompt the user)
- Else:
- Tell the user you couldn’t auto-install git on this OS and show the failing detection output.
Re-check:
git --version
Only notify the user if install failed.
1B) Ensure cron/crontab is available (best-effort auto-install)
Check:
command -v crontab
If missing, attempt install:
- If
apt-getexists:sudo apt-get update && sudo apt-get install -y cronsudo systemctl enable --now cron || sudo service cron start || true
- Else if
dnfexists:sudo dnf install -y croniesudo systemctl enable --now crond || true
- Else if
yumexists:sudo yum install -y croniesudo systemctl enable --now crond || true
- Else if
pacmanexists:sudo pacman -S --noconfirm croniesudo systemctl enable --now cronie || true
- Else if
apkexists:sudo apk add dcronsudo rc-update add dcron default || truesudo rc-service dcron start || true
- Else:
- If you can’t install, tell the user cron is required for scheduling.
Re-check:
command -v crontab
Step 2: Ensure GitHub CLI (gh) is installed (auto-install)
Check:
gh --version
If missing, install:
-
If
brewexists:brew install gh
-
Else if
apt-getexists (official GitHub CLI packages; preferred):- Install using the official apt repo steps:
(type -p wget >/dev/null || (sudo apt-get update && sudo apt-get install -y wget))sudo mkdir -p -m 755 /etc/apt/keyringsout=$(mktemp) && wget -nv -O"$out" https://cli.github.com/packages/githubcli-archive-keyring.gpgcat "$out" | sudo tee /etc/apt/keyrings/githubcli-archive-keyring.gpg > /dev/nullsudo chmod go+r /etc/apt/keyrings/githubcli-archive-keyring.gpgsudo mkdir -p -m 755 /etc/apt/sources.list.decho "deb [arch=$(dpkg --print-architecture) signed-by=/etc/apt/keyrings/githubcli-archive-keyring.gpg] https://cli.github.com/packages stable main" | sudo tee /etc/apt/sources.list.d/github-cli.list > /dev/nullsudo apt-get update && sudo apt-get install -y gh
- Install using the official apt repo steps:
-
Else if
dnfexists:sudo dnf install -y 'dnf-command(config-manager)' || sudo dnf install -y dnf5-plugins || truesudo dnf config-manager --add-repo https://cli.github.com/packages/rpm/gh-cli.repo || sudo dnf config-manager addrepo --from-repofile=https://cli.github.com/packages/rpm/gh-cli.repo || truesudo dnf install -y gh --repo gh-cli || sudo dnf install -y gh || true
-
Else if
yumexists:type -p yum-config-manager >/dev/null || sudo yum install -y yum-utilssudo yum-config-manager --add-repo https://cli.github.com/packages/rpm/gh-cli.reposudo yum install -y gh
-
Else if
zypperexists:sudo zypper addrepo https://cli.github.com/packages/rpm/gh-cli.repo || truesudo zypper refsudo zypper install -y gh
-
Else if
pacmanexists:sudo pacman -S --noconfirm github-cli
-
Else if
apkexists:sudo apk add github-cli
-
Else:
- Tell the user you can’t auto-install
ghon this OS.
- Tell the user you can’t auto-install
Re-check:
gh --version
Only notify the user if install failed.
Step 3: Ensure the user is authenticated in gh (agent runs the flow)
Check:
gh auth status --hostname github.com
If NOT authenticated:
-
Run:
gh auth login --hostname github.com --git-protocol https
-
The terminal flow will show a one-time code and ask the user to authorize.
- Tell the user to open https://github.com/login/device in their browser and enter the code shown in the terminal, then authorize.
-
After login:
gh auth setup-git
-
Verify again:
gh auth status --hostname github.com
If auth fails, stop and report the exact terminal output.
Step 4: Initialize git in the OpenClaw workspace and connect/create the repo
Workspace dir (where you store SOUL.md, AGENTS.md, etc.):
- Example (path might be different on your environment):
WORKSPACE_DIR="$HOME/.openclaw/workspace"
-
Ensure the workspace exists:
mkdir -p "$WORKSPACE_DIR"cd "$WORKSPACE_DIR"
-
Initialize repo if needed:
- If
.gitdoes not exist:git init git branch -M main
- If
-
Configure a deterministic commit identity (local-only):
git config user.name "gitclaw.ai"git config user.email "gitclaw-bot@users.noreply.github.com"
-
Determine the authenticated GitHub username (owner):
OWNER="$(gh api user --jq .login)"- (Do not print unless debugging is needed)
-
Repo name and visibility:
REPO="<repo name from user>"- Visibility:
public=>--publicprivate=>--private
-
Ensure there is at least one commit (required for first push/cron):
- Create a tiny marker file if needed:
test -f .gitclaw.keep || printf "gitclaw initialized: %s\n" "$(date -u '+%Y-%m-%dT%H:%M:%SZ')" > .gitclaw.keep
git add -Agit commit -m "gitclaw: initial backup" || true
- Create a tiny marker file if needed:
-
Create or reuse the target repo:
- If it exists:
gh repo view "$OWNER/$REPO" >/dev/null 2>&1- Set remote:
REMOTE_URL="https://github.com/$OWNER/$REPO.git"- If origin exists:
git remote set-url origin "$REMOTE_URL" - Else:
git remote add origin "$REMOTE_URL"
- Try to fast-forward sync (avoid overwriting remote history):
git fetch origin main || truegit merge --ff-only origin/main || true
- If it does NOT exist:
- Create it non-interactively and connect it:
- Public:
gh repo create "$REPO" --public --confirm
- Private:
gh repo create "$REPO" --private --confirm
- Public:
- Set remote:
REMOTE_URL="https://github.com/$OWNER/$REPO.git"git remote add origin "$REMOTE_URL" || git remote set-url origin "$REMOTE_URL"
- Create it non-interactively and connect it:
- If it exists:
-
Initial push:
git push -u origin main
If push fails due to conflicts or non-fast-forward:
- Do NOT force-push automatically.
- Report the exact error and stop (user decision required).
Step 5: Install deterministic backup script (NO AI / NO heartbeat)
Create a folder outside the workspace:
mkdir -p "$HOME/.openclaw/gitclaw"
Create this script EXACTLY:
Path:
$HOME/.openclaw/gitclaw/auto_backup.sh
Contents:
#!/usr/bin/env bash
set -euo pipefail
# GitClaw deterministic backup (no AI)
export PATH="/usr/local/bin:/opt/homebrew/bin:/usr/bin:/bin:$PATH"
WORKSPACE_DIR="${HOME}/.openclaw/workspace"
STATE_DIR="${HOME}/.openclaw/gitclaw"
LOG_FILE="${STATE_DIR}/backup.log"
LOCK_DIR="${STATE_DIR}/lock"
mkdir -p "${STATE_DIR}"
timestamp() { date -u '+%Y-%m-%dT%H:%M:%SZ'; }
# Simple lock to prevent overlapping runs
if ! mkdir "${LOCK_DIR}" 2>/dev/null; then
echo "$(timestamp) Skip: already running." >> "${LOG_FILE}"
exit 0
fi
trap 'rmdir "${LOCK_DIR}" >/dev/null 2>&1 || true' EXIT
if ! command -v git >/dev/null 2>&1; then
echo "$(timestamp) ERROR: git not found on PATH. Install git first." >> "${LOG_FILE}"
exit 2
fi
if [ ! -d "${WORKSPACE_DIR}/.git" ]; then
echo "$(timestamp) ERROR: ${WORKSPACE_DIR} is not a git repo. Run GitClaw setup first." >> "${LOG_FILE}"
exit 3
fi
cd "${WORKSPACE_DIR}"
# Stage everything
git add -A
# If nothing staged, exit quietly
if git diff --cached --quiet; then
echo "$(timestamp) No changes." >> "${LOG_FILE}"
exit 0
fi
# Commit + push
git commit -m "gitclaw backup: $(timestamp)" >> "${LOG_FILE}" 2>&1
git push origin main >> "${LOG_FILE}" 2>&1
echo "$(timestamp) Backup OK." >> "${LOG_FILE}"
Write the script to:
$HOME/.openclaw/gitclaw/auto_backup.sh
Then:
chmod +x "$HOME/.openclaw/gitclaw/auto_backup.sh"
Step 6: Configure crontab (idempotent)
Default schedule: hourly (0 * * * *). If user provided a different frequency, convert it to a cron expression.
- Define:
CRON_CMD="$HOME/.openclaw/gitclaw/auto_backup.sh"CRON_LINE="0 * * * * $CRON_CMD"
- Install without duplicates:
crontab -l 2>/dev/null | grep -F "$CRON_CMD" >/dev/null- If not found, append:
(crontab -l 2>/dev/null; echo "$CRON_LINE") | crontab -
- Confirm:
crontab -l | grep -F "$CRON_CMD"
Step 7: Final validation
- Run once:
$HOME/.openclaw/gitclaw/auto_backup.sh
- Show the log:
tail -n 50 "$HOME/.openclaw/gitclaw/backup.log" || true
- Tell the user:
- Repo:
https://github.com/$OWNER/$REPO - Schedule: hourly (or the chosen cadence)
- Script path:
~/.openclaw/gitclaw/auto_backup.sh
Top skills in this category
Free Ride - Unlimited free AI
@shaivpidadiManages free AI models from OpenRouter for OpenClaw. Automatically ranks models by quality, configures fallbacks for rate-limit handling, and updates opencla...
Playwright (Automation + MCP + Scraper)
@ivangdavilaAutomates, tests, and debugs browsers with Playwright: locators, auto-waiting, traces, CI runs, and MCP browser control. Use when a test is flaky, times out, or fails only in CI or headless; when a locator matches multiple elements or the wrong one (strict mode violation); when clicks need force, waits become sleeps, or networkidle never settles; for storageState and login setup, request mocking and HAR replay, uploads and downloads, iframes and shadow DOM, popups and dialogs, screenshot diffs that change per machine, trace and report artifacts, sharding a slow suite, device and permission emulation, accessibility checks, driving a real browser through Playwright MCP, extracting data from JS-rendered pages, or porting a Cypress, Puppeteer, or Selenium suite to Playwright. Not for maintaining an existing Cypress or Puppeteer suite (cypress, puppeteer) or for work a plain HTTP request answers (http).
n8n
@thomasansemsManage n8n workflows and automations via API. Use when working with n8n workflows, executions, or automation tasks - listing workflows, activating/deactivating, checking execution status, manually triggering workflows, or debugging automation issues.
Superpowers Dev Workflow
@wlshlad85Spec-first, TDD, subagent-driven software development workflow. Use when: (1) building any new feature or app — triggers brainstorm → plan → subagent executi...
Google Ads
@jdrhyneQuery, audit, and optimize Google Ads campaigns. Supports two modes: (1) API mode for bulk operations with google-ads Python SDK, (2) Browser automation mode for users without API access - just attach a browser tab to ads.google.com. Use when asked to check ad performance, pause campaigns/keywords, find wasted spend, audit conversion tracking, or optimize Google Ads accounts.