ClawHub

WebChat Image Support

v1.0.0

Enables agents to detect and analyze images sent via WebChat or other channels using vision-capable models or fallback media processing.

⭐ 1· 73·0 current·0 all-time
by@styoha
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 match the instructions: the SKILL.md explains detecting images, selecting vision-capable models, and falling back to the media pipeline. No unrelated credentials, binaries, or capabilities are requested.
ℹ
Instruction Scope
The instructions reference a specific local config path (~/.openclaw/agents/main/agent/models.json) and tell operators how to configure vision-capable models. They do not instruct the agent to read or exfiltrate arbitrary files or credentials, but they do assume the operator or agent has access to and may edit that config.
✓
Install Mechanism
No install spec or code files are present (instruction-only), so nothing is downloaded or written to disk by the skill itself.
✓
Credentials
The skill requests no environment variables, credentials, or config paths beyond recommending edits to the models.json config required for vision models; this is proportionate to an image-support feature.
✓
Persistence & Privilege
The skill is not always-enabled and does not request persistent system-wide modifications. It contains no code that would run autonomously outside normal agent invocation.
Assessment
This is an instruction-only skill that tells you how to enable image support by configuring your OpenClaw gateway and models.json. Before installing/using it, verify your OpenClaw version meets the requirement (2026.3.29+), confirm you have at least one vision-capable model configured, and ensure you (or your admin) are comfortable editing ~/.openclaw/agents/main/agent/models.json. The skill does not request credentials or install code, but it assumes the platform's media pipeline and configured models are trusted. If you need automated changes to config files, prefer doing that manually or through your standard provisioning process rather than granting any third party write access.

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

latestvk97emns4wxwmd5cawcb0hnc2ss83vh92

License

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

Comments