ClawHub

OpenClaw Model Optimizer

v1.1.0

Optimize OpenClaw model configuration by declaring missing model capabilities (vision/multimodal input, context window, max output tokens, reasoning). Use wh...

⭐ 1· 97·0 current·0 all-time
by@yay2008
MIT-0
Download zip
LicenseMIT-0 · Free to use, modify, and redistribute. No attribution required.
Security Scan
VirusTotalVirusTotal
Benign
View report →
OpenClawOpenClaw
Benign
high confidence
✓
Purpose & Capability
Name/description (model capability declarations for OpenClaw) match the actions described: reading and updating openclaw.json to add agents.defaults.models entries. No unrelated credentials, binaries, or external services are required.
ℹ
Instruction Scope
Runtime instructions explicitly read and write the user's OpenClaw config (~/.openclaw/openclaw.json) and create a backup. The skill instructs the agent to ask for user confirmation before applying changes and not to overwrite existing declarations without consent — this is appropriate but does mean the agent will need filesystem read/write access to that path.
✓
Install Mechanism
Instruction-only skill with no install spec and no code files. Nothing is downloaded or written by an installer, minimizing installation risk.
✓
Credentials
No environment variables, credentials, or config paths beyond the single OpenClaw config file are requested. The requested access is proportional to the stated task.
ℹ
Persistence & Privilege
always:false and no special privileges are requested. The only persistence is modifying a local config file (with a backup). Users should be aware the skill will alter a user config file if they confirm the changes.
Assessment
This skill is coherent for its purpose but will read and modify your OpenClaw config file (~/.openclaw/openclaw.json). It promises to create a backup and to ask for confirmation before making changes — verify the backup exists before proceeding, carefully review any capability values it proposes (provider-id/model-id keys must exactly match your config), and restart OpenClaw after edits. Because it's an instruction-only skill with no external installs or credential requests, there is no obvious exfiltration risk from the skill itself, but always review proposed edits before allowing them.

Like a lobster shell, security has layers — review code before you run it.

latestvk974nt23rd9m8ryfr94sbjm8vx83cg6j

License

MIT-0
Free to use, modify, and redistribute. No attribution required.
Termshttps://spdx.org/licenses/MIT-0.html

Comments