Switch Modes
ReviewAudited by ClawScan on May 10, 2026.
Overview
The skill is mostly coherent, but it persistently changes OpenClaw’s global model configuration and its documentation gives conflicting scope expectations about whether the change affects only the current session or all sessions.
Install only if you want a skill that can change OpenClaw's default model by editing ~/.openclaw/openclaw.json. Be aware this may affect other sessions and future requests, not just the current chat, and verify the config file is backed up or easily restorable before use.
Findings (3)
Artifact-based informational review of SKILL.md, metadata, install specs, static scan signals, and capability signals. ClawScan does not execute the skill or run runtime probes.
A short command can change which model future OpenClaw requests use, potentially affecting performance, behavior, and API costs.
The skill directly changes the agent's main model configuration. That is purpose-aligned, but it is a high-impact mutation of agent behavior and cost profile.
When user requests a mode switch: ... Read current config: `~/.openclaw/openclaw.json` ... Update the `model` field with the new model ID ... Write back to `~/.openclaw/openclaw.json`
Before installing, confirm you are comfortable with this skill changing OpenClaw's main config. Prefer requiring explicit confirmation before switching to expensive modes such as SMART or MAX.
A mode switch in one conversation may unexpectedly affect other OpenClaw sessions or future work.
The model change persists globally across sessions. This conflicts with SKILL.md's note that changes 'only affect the current session's default model,' creating unclear user expectations about blast radius.
Does this affect other OpenClaw sessions? A: Yes. It modifies the global OpenClaw config, so all sessions use the new default model.
The skill should clearly state that it changes the global OpenClaw default model and should offer an undo or session-only option if supported.
If implemented carelessly by the agent, the config rewrite could expose or corrupt API key settings.
The referenced OpenClaw config file may contain provider API keys. The skill's instructions say to preserve other settings and only modify the model field, and there is no evidence of exfiltration, but users should recognize it handles a credential-bearing file.
"apiKeys": { "anthropic": "sk-...", "openai": "sk-..." }Ensure the agent never displays API keys and only rewrites the `model` field while preserving the rest of the file exactly.