← Back to Skills Marketplace
292
Downloads
1
Stars
0
Active Installs
4
Versions
Install in OpenClaw
/install openclaw-session-model-switcher
Description
Instantly switch the current OpenClaw session model; supports gpt, claude, qianwen, minimax, current model status, configured model listing, and restoring th...
Usage Guidance
This skill is plausible for switching session models, but it is incomplete and relies on external scripts and the node runtime that are not included or declared. Before installing or running it: 1) Verify that scripts/switch-model.sh, scripts/list-models.sh, and scripts/model-status.sh exist in your OpenClaw workspace and inspect their contents—handler.sh will chmod and execute them. 2) Ensure node is available on PATH (handler.sh uses node). 3) Understand that this package prints session commands for you to run rather than performing the switch itself in many cases (SKILL.md suggests direct execution, but handler.sh mostly prints commands). 4) If you cannot inspect or control the external scripts, do not install or run this skill in a sensitive environment. If the author intended a bundled, self-contained skill, ask them to include or document the external scripts and declare required binaries.
Capability Analysis
Type: OpenClaw Skill
Name: openclaw-session-model-switcher
Version: 0.1.3
The skill is designed to manage AI model switching within an OpenClaw session by interacting with system scripts (switch-model.sh, list-models.sh, model-status.sh). While handler.sh performs a chmod +x on scripts outside its own directory, this appears to be a functional requirement for the environment rather than a malicious act. The logic for parsing user input and handling model aliases in model-aliases.json is consistent with the stated purpose, and no evidence of data exfiltration or unauthorized execution was found.
Capability Assessment
Purpose & Capability
The skill claims to switch the current session model, list configured models, and report status. Its bundled files (handler.sh, model-aliases.json, README, SKILL.md) are consistent with that purpose, but handler.sh expects workspace scripts (scripts/switch-model.sh, scripts/list-models.sh, scripts/model-status.sh) that are not included in the package. It also invokes node for small JS snippets but the skill declares no required binaries. Requiring external, out-of-package scripts and an undeclared runtime is disproportionate to the stated self-contained capability.
Instruction Scope
SKILL.md instructs the agent to read models.providers from the active OpenClaw configuration and, when possible, directly execute session-level commands. The provided handler.sh, however, does not read OpenClaw config directly — it delegates to external scripts and prints commands rather than executing them. handler.sh also chmods and executes scripts found under a workspace path (../..). Delegating to and executing external workspace scripts broadens scope beyond the skill bundle and could execute arbitrary code depending on those scripts' contents.
Install Mechanism
There is no install specification (instruction-only with a small handler script). No network downloads or package installs are declared. This is low-risk in terms of install mechanism, but the lack of an install step combined with reliance on out-of-package scripts means runtime behavior depends on the user's environment.
Credentials
The skill declares no required environment variables or credentials, which is appropriate. However, handler.sh depends on the node binary and on writable/executable workspace scripts (it runs chmod +x and executes them) without declaring these requirements. That undeclared dependency (node) and implicit need to modify script permissions is a proportionality concern and should be documented.
Persistence & Privilege
The skill does not request permanent presence (always: false), does not claim to modify global configs, and the files do not modify other skills. The only filesystem action in the bundle is chmod on workspace scripts (if present), which affects files outside the skill directory — noteworthy but not an explicit persistence/privilege escalation by itself.
How to Use
- Make sure OpenClaw is installed (local or Docker)
- Run the install command in chat:
/install openclaw-session-model-switcher - After installation, invoke the skill by name or use
/openclaw-session-model-switcher - Provide required inputs per the skill's parameter spec and get structured output
Version History
v0.1.3
Polish update: translate handler output to English while preserving multilingual input matching and session-level model switching behavior.
v0.1.2
Maintenance update: align local folder naming and internal file references with the published OpenClaw Session Model Switcher slug.
v0.1.1
English docs refresh: translate skill metadata, summary, and README; keep session-level switching behavior unchanged.
v0.1.0
Initial release: manually switch the current OpenClaw session model, list configured models, check model status, and restore default without editing global config or restarting Gateway.
Metadata
Frequently Asked Questions
What is OpenClaw Session Model Switcher?
Instantly switch the current OpenClaw session model; supports gpt, claude, qianwen, minimax, current model status, configured model listing, and restoring th... It is an AI Agent Skill for Claude Code / OpenClaw, with 292 downloads so far.
How do I install OpenClaw Session Model Switcher?
Run "/install openclaw-session-model-switcher" in the OpenClaw or Claude Code chat to install it in one step — no extra setup required.
Is OpenClaw Session Model Switcher free?
Yes, OpenClaw Session Model Switcher is completely free, licensed under MIT-0. You can download, install and use it at no cost.
Which platforms does OpenClaw Session Model Switcher support?
OpenClaw Session Model Switcher is cross-platform and runs anywhere OpenClaw / Claude Code is available (cross-platform).
Who created OpenClaw Session Model Switcher?
It is built and maintained by dpbs (@dpbs-715); the current version is v0.1.3.
More Skills