Ezviz Open Restaurant Inspection
ReviewAudited by ClawScan on May 10, 2026.
Overview
The skill matches its Ezviz camera-inspection purpose, but it caches powerful Ezviz access tokens in a shared temporary location with weak write-time protection, so it should be reviewed before use.
Use only dedicated minimal-permission Ezviz credentials, confirm the configured cameras are intended, and consider disabling token caching unless the cache implementation is fixed. Review the bundled Python files and make sure you are comfortable with images being sent to Ezviz AI and retained for two hours.
Findings (5)
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.
On a shared system, another local process or same-user skill may be able to read or race the token cache and use the Ezviz token until it expires.
The cache stores Ezviz access tokens and writes them to a predictable /tmp cache file using default file creation permissions before applying chmod 0600. The cache directory is also not shown as mode-restricted.
all_cache[cache_key] = {"access_token": access_token, ...}; with open(temp_file, 'w') as f: json.dump(cache_data, f, indent=2); os.replace(temp_file, cache_file); os.chmod(cache_file, 0o600)Disable caching with EZVIZ_TOKEN_CACHE=0, or update the cache code to create the directory with 0700 permissions and create temporary files with 0600 permissions before writing any token data.
If local OpenClaw config files already contain Ezviz credentials, the skill may use those stored credentials when run.
The skill can fall back to local OpenClaw configuration files to load Ezviz app credentials and devices, not only environment variables.
CONFIG_PATHS = [Path.home() / ".openclaw" / "config.json", Path.home() / ".openclaw" / "gateway" / "config.json", Path.home() / ".openclaw" / "channels.json"]
Use dedicated, minimal-permission Ezviz app credentials and verify the local OpenClaw config files before running the skill.
Running the skill can change the Ezviz account state by creating an intelligent agent and can capture images from configured cameras.
The skill discloses high-impact remote API actions, including account-side agent creation and camera image capture.
Running this skill means you accept these remote actions: ... Create agent from template if needed ... Capture device images ... Send images for AI analysis
Run it only after explicit approval, test with non-production devices first, and confirm the listed device serials are the intended cameras.
Camera images may contain private restaurant, employee, customer, or operational information and will leave the local environment.
The documentation clearly states that captured device images are sent to Ezviz services for AI analysis and retained temporarily.
Data flow: Device -> open.ys7.com -> aidialoggw.ys7.com -> Local output ... Images stored on Ezviz (2 hours)
Use the skill only for cameras and locations where this data transfer and retention period are acceptable.
Users have less external context for who maintains the skill before granting it camera and Ezviz API access.
The registry metadata does not provide an upstream source or homepage for provenance checking.
Source: unknown; Homepage: none
Review the bundled scripts and verify the intended Ezviz API domains before use.