Browse / General-purpose / CraigClaw
CraigClaw
HashGrid-Networks
Craig's OpenClaw workspace — personality, memory, skills, and deployment for Merit-Systems' AI agent
Open HashGrid-Networks/CraigClaw → Does not match the four rules.
Works with any agent
ships AGENTS.md, the cross-tool instruction file.
In its own words
Craig2 is Merit-Systems' AI agent, powered by OpenClaw. He lives on Discord, has write access to every repo in the Merit-Systems org (but can't merge), and creates PRs directly. This repo IS Craig's brain -- his personality, memory, skills, and operating instructions are all here, version-controlled and auto-deployed.
How the Workspace-as-Repo Sync Works
This is not default OpenClaw behavior. By default, OpenClaw loads workspace files (SOUL.md, AGENTS.md, etc.) from a local directory and that's it. We added a version-controlled sync layer on top:
- Inbound (deploy): A human pushes to
main(or a PR is merged) → GitHub Actions SSHes into EC2 → runsgit pull→ restarts the OpenClaw gateway → Craig's next session loads the updated files. This is how humans edit Craig's brain.
- Outbound (self-edit): Craig edits a file in his workspace on EC2 → creates a branch, commits, pushes → creates a PR via
ghCLI → a human reviews and merges → triggers the inbound deploy. This is how Craig edits his own brain.
The key insight: OpenClaw's workspace directory is just a git clone of this repo. By keeping them in sync via GitHub Actions, every change (whether from Craig or a human) is version-controlled, reviewable, and auto-deployed.
Quoting 16% of the README. Read the rest on GitHub ↗
This one grew out of another entry. It began as a copy of MadsLorentzen/ai-job-search, and most of its files are now different from the original, so it is listed in its own right rather than behind its source.
Where it came from
What this copy built: +deploy +memory +skills — top-level directories this copy added on top of the original.
Read from the repository's git tree on 2026-08-17.