The gateway failed with ‘missing env var.’ I had the token—just under the old project name.

My Hugo blog showed zero posts. The content was there—just pointed at a user who doesn’t exist on this machine.

Learn how to systematically migrate legacy naming across a complex codebase and ensure they never return using automated enforcement.

Renaming a project sounds trivial until you discover how many hidden artifacts still reference the old name. A case study in env var migration pain.

What Happened

OpenClaw gateway failed to start with:

MissingEnvVarError: Missing env var "CLAWDBOT_GATEWAY_TOKEN" 
referenced at config path: gateway.auth.token

User had the token set, but under wrong name.

Investigation

Project was renamed: clawdbot → moltbot → openclaw

Config file (~/.clawdbot/moltbot.json) still referenced old env var:

"gateway": {
  "auth": {
    "token": "${CLAWDBOT_GATEWAY_TOKEN}"
  }
}

But the new convention uses OPENCLAW_* prefix.

Resolution

1. Updated config to use ${OPENCLAW_GATEWAY_TOKEN}
2. Renamed env var in Keychain and wrapper script
3. Doctor command now warns about deprecated vars:

- Deprecated legacy environment variables detected (ignored).
- Use OPENCLAW_* equivalents instead:
  CLAWDBOT_GATEWAY_TOKEN -> OPENCLAW_GATEWAY_TOKEN

Lesson

Renaming projects is deceptively hard. Things that need updating:

OpenClaw’s doctor command handles some of this via auto-migration, but env vars in user scripts/configs require manual updates.

What Happened

Hugo blog had a symlink content/posts pointing to /Users/deok/brainFucked/10-Blog/published - but the current user is koed, not deok.

The blog appeared set up but couldn’t find any content. The path was a remnant from a different machine/user setup.

Why It Matters

Common gotcha when:

• Migrating dotfiles/configs between machines
• Cloning repos with absolute path symlinks
• Working across different user accounts

Symlinks with hardcoded absolute paths don’t survive machine migrations.

Resolution

rm content/posts
ln -s /Users/koed/Dev/BrainFucked/10-Blog/published content/posts

Also had to create the published/ directory which didn’t exist.

Lesson

• Prefer relative symlinks when possible
• Check symlink targets after cloning repos
• Audit absolute paths in dotfiles before migration