Query and monitor UniFi network via local gateway API (Cloud Gateway Max / UniFi OS). Use when the user asks to "check UniFi", "list UniFi devices", "show who's on the network", "UniFi clients", "UniFi health", "top apps", "network alerts", "UniFi DPI", or mentions UniFi monitoring/status/dashboard.
Security Analysis
high confidenceThe skill mostly does what it claims (read-only UniFi monitoring), but there are some mismatches and risky implementation details (undeclared config file, plaintext credentials, files written to disk, and disabled SSL verification) that you should review before installing.
The scripts and documentation align with the stated purpose: they use the UniFi OS local gateway API to GET monitoring endpoints (devices, clients, health, DPI, alerts). Required binaries (curl, jq) are appropriate. However, the skill expects a credentials file at ~/.clawdbot/credentials/unifi/config.json (or environment variables) even though the registry metadata declared no required config paths or env — that metadata omission is an incoherence.
SKILL.md and the scripts instruct the agent to read a local credentials file (containing username/password) and to perform login (POST /api/auth/login) then multiple GETs. The scripts also write data to disk (dashboard_debug_dump.json and $HOME/clawd/memory/bank/unifi-inventory.md). The README and scripts intentionally skip SSL verification (curl -k). While the operations are limited to the local UniFi gateway and are GET-only for configuration endpoints, the skill stores and reads plaintext credentials and persists potentially sensitive network data to files — these behaviors extend beyond simple transient monitoring and increase risk if not managed.
Instruction-only skill with bundled shell scripts (no external downloads or package installs). No install spec means nothing arbitrary is fetched during install — low install risk. The included scripts will be executed at runtime, so their contents matter (and were reviewed).
Requesting UniFi local admin credentials is expected for the stated purpose, but the skill did not declare the required config path or environment variables in the registry metadata. Storing a local admin password in plaintext at ~/.clawdbot/credentials/unifi/config.json (or via env vars) is necessary for this implementation but is sensitive — the skill doesn't provide guidance on secure file permissions beyond the README. Using UNIFI_CONFIG_FILE override is supported but again not declared in metadata.
always:false and autonomous invocation not disabled (normal). The scripts create and write files under the user's home (OUTPUT_FILE=$HOME/clawd/memory/bank/unifi-inventory.md and dashboard_debug_dump.json). They also create temporary cookie files for sessions (deleted in most flows). Writing persistent dumps of network inventory/clients to disk is legitimate for some workflows but is a persistence/privacy concern that the user should explicitly accept.
Guidance
This skill appears to implement a legitimate UniFi read-only monitor, but review and accept a few risks before installing: - Credentials: the skill requires a local UniFi admin username/password stored in ~/.clawdbot/credentials/unifi/config.json (or via environment variables). That file contains plaintext credentials — ensure you set restrictive permissions (chmod 600) and consider creating a dedicated low-privilege local admin for the API. - Metadata mismatch: the registry entry lists no required config paths/env, but the skill actually requires the config file or env vars. Treat this as an indicator the package metadata is incomplete; verify the file location and ownership before running. - Disk writes: the scripts write files (dashboard_debug_dump.json and $HOME/clawd/memory/bank/unifi-inventory.md) that contain sensitive network data. If you don’t want that persisted, edit the scripts to disable dumps or change paths to a secure location. - SSL verification: the scripts use curl -k to skip TLS verification (common with self-signed UniFi certs). This is insecure in network environments where the gateway might be spoofed — if possible, use a valid certificate or modify scripts to verify certs. - Review scripts locally: because this is an instruction-based skill with runnable shell scripts, inspect scripts locally before use and consider running them in a controlled environment first. If you’re uncomfortable with plaintext creds or disk persistence, either modify the scripts to use a more secure auth method (token, restricted account) or do not install. If you want, I can list the exact places in the scripts to change to improve security (e.g., remove debug dump, enforce cert verification, set strict file perms).
Latest Release
v1.0.1
- Added a new README.md file with documentation. - Updated SKILL.md to increment version to 1.0.1. - No functional changes to code or features.
More by @jmagar
Published by @jmagar on ClawHub
