Apothem vs raw Open-Claw config
Hand-maintaining Open-Claw config versus managing it through Apothem's shared profile.
Open-Claw reads ~/.openclaw/openclaw.json. Its agents.defaults.skills is a
name allowlist (not a directory loader) and MCP is a CLI surface, so Apothem
authors no config keys; it installs shared command and skill cohorts under
~/.openclaw/.apothem/support/skills/ as support material.
Status
Preview. The Open-Claw adapter targets the current documented config file and extra skill-directory surface. Non-skill cohorts remain Apothem support material.
Effort comparison
| Task | Raw hand-config | Apothem-managed |
|---|---|---|
| Author reusable skills | Hand-edit config and skill directories | Write once in the shared corpus |
| Keep cohorts aligned with other harnesses | Manual, per tool | Automatic on apothem update |
| Install / remove the config | Manual | apothem install / apothem uninstall |
| Verify the config is correct | Manual inspection | apothem verify --harness open-claw |
What Apothem manages
The Open-Claw adapter writes openclaw.json, wraps Apothem slash commands as
skills, copies Apothem skills into the referenced support directory, and keeps
rules, templates, agents, and hook material under the Apothem support subtree.
Migration path
- Run
apothem profile initto create the shared profile if you have not yet. - Move reusable Open-Claw skills or command prompts into the shared corpus.
- Run
apothem install --harness open-clawto render the native config and support skill directory. - Run
apothem verify --harness open-clawto confirm the config landed.