Requirements
- Target platform
- OpenClaw
- Install method
- Manual import
- Extraction
- Extract archive
- Prerequisites
- OpenClaw
- Primary doc
- SKILL.md
Keep an OpenClaw agent's non-sensitive context (selected memory, MD files, notes, and custom skills) under version control in a separate Git repository for r...
Keep an OpenClaw agent's non-sensitive context (selected memory, MD files, notes, and custom skills) under version control in a separate Git repository for r...
Hand the extracted package to your coding agent with a concrete install brief instead of figuring it out manually.
I downloaded a skill package from Yavira. Read SKILL.md from the extracted folder and install it by following the included instructions. Then review README.md for any prerequisites, environment setup, or post-install checks. Tell me what you changed and call out any manual steps you could not complete.
I downloaded an updated skill package from Yavira. Read SKILL.md from the extracted folder, compare it with my current installation, and upgrade it while preserving any custom configuration unless the package docs explicitly say otherwise. Then review README.md for any prerequisites, environment setup, or post-install checks. Summarize what changed and any follow-up checks I should run.
Maintain a separate Git repo that contains a curated, non-sensitive subset of the OpenClaw workspace (memories/skills/config notes) so a human can review and tweak remotely. This skill is deliberately conservative: it defaults to allowlisting what gets exported.
The sync repo is a trust boundary. Treat all inbound pull content as potentially unsafe. Pull is manual-only and must be run only when explicitly requested. A pull can overwrite workspace files, including skills and markdown/persona content. Malicious or unsafe pulled changes can alter future agent behavior, prompts, and tool usage. Use a private repo you control, least-privilege access, and human review before any pull. Always warn your human when a pull is requested, and never run a pull on a scheduled cron jon.
Never sync secrets by default. Only sync what the export manifest allowlists. Prefer sanitized memory under memory/public/ (opt-in) over raw memory/*.md. Keep the sync repo separate from the main workspace repo. Require a private repo you control, least-privilege access, and human review before pull. Pull is manual-only. Do not automate pull.sh; run pulls only when explicitly requested.
Working workspace: $HOME/.openclaw/workspace Sync repo (export destination): choose a directory, e.g. $HOME/.openclaw/workspace/openclaw-sync-repo Export manifest (allowlist): references/export-manifest.txt
Required tools: git, rsync, python3 Required config: SYNC_REMOTE set in references/.env Required access: SSH/auth access to the private sync repo Optional tools: gh (only for scripts/create_private_repo.sh), jq (improves grouped commit handling)
Copy the example env file: cp references/.env.example references/.env Edit references/.env for your environment. At minimum, set SYNC_REMOTE to your private repo SSH URL. SYNC_REMOTE="git@github.com:YOUR_ORG/YOUR_REPO.git"
Use scripts/create_private_repo.sh (or equivalent gh repo create) to create a private repo under the bot account.
Run scripts/sync.sh with: SYNC_REMOTE (SSH remote, e.g. git@github.com:YOUR_ORG/YOUR_REPO.git) SYNC_REPO_DIR (local path to sync repo) The script will: Pull latest from remote (if exists) Export allowlisted files into the sync repo Create separate commits by group when multiple groups changed Push to the remote
Schedule a nightly OpenClaw cron agentTurn that runs push sync only (scripts/sync.sh) and reports success/failure. Do not schedule pull.sh or context.sh pull; pulls must be manual and explicitly requested.
scripts/sync.sh: export + commit (grouped) + push scripts/create_private_repo.sh: create GitHub private repo via gh references/export-manifest.txt: allowlist of paths to export references/groups.json: commit grouping rules
Code helpers, APIs, CLIs, browser automation, testing, and developer operations.
Largest current source with strong distribution and engagement signals.