MCP App StoreMCP App Store
from deepcamera2,688
Real-time monocular depth estimation for camera feeds with a privacy-focused depth-only mode to anonymize people while preserving scene layout.
This skill provides real-time monocular depth estimation using Depth Anything v2, producing colorized depth maps and overlays for live camera feeds. It supports multiple backends (CoreML on macOS with Apple Neural Engine and PyTorch on Linux/Windows with CUDA/CPU) and includes a privacy-oriented depth_only blend mode that removes identifying visual features while preserving spatial layout and activity for monitoring and analytics. The skill exposes a TransformSkillBase-compatible interface so it can be integrated into Aegis-style camera pipelines and other video processing systems.
Use this skill when you need depth-aware video transforms: privacy-preserving monitoring (hide identities but keep movement/positions), depth overlays for visualization, or 3D scene understanding for downstream analytics (people counting, proximity alerts, occlusion reasoning). It's suitable for edge devices (Apple Neural Engine acceleration) and server/desktop deployments with CUDA.
Best with agents or systems that integrate with camera transform skills and Python-based pipelines (agents that can run local Python scripts, CoreML-enabled macOS agents, or vision-focused Copilot/Code agents).
The SKILL.md describes a 'Depth Anything v2' depth estimation privacy skill, but 6 of the 7 bundled scripts originate from an unrelated face recognition project (model-r50-am-lfw, src/embedding, src/face_detection). Only transform.py matches the stated purpose. This mismatch between advertised function and bundled scripts is a significant concern. The scripts have no error handling, use deprecated pip2, download binary blobs from GitHub without checksum verification, and start sshd/redis/mosquitto as background services — none of which relate to depth estimation.
transform_basedepth_anything_v2coremltoolsScripts appear to be from github.com/solderzzc face-detection project accidentally (or intentionally) bundled into a depth estimation skill. Not clearly malicious, but the severe content mismatch warrants manual review. Security score held at 7 because wget-only (no pipe-to-shell), no hardcoded credentials, but sshd startup and unverified binary downloads reduce confidence.
AWP (Agent Work Protocol)
Tooling and scripts for onboarding, staking, allocation, and managing agents on the AWP network (Base/Ethereum/Arbitrum/BSC). Includes safe, opt-in daemon and r
Cost Tracker
Monitor agent session costs, set budget alerts, and get actionable token-spend optimizations to keep multi-session workflows within budget.
Canary — Post-Deploy Visual Monitor
Run a short post-deploy monitor that captures screenshots, checks console errors, and compares performance against baselines to detect regressions and page fail
Monitoring Stack Deployer
Deploy and configure production-ready monitoring stacks (Prometheus, Grafana, Datadog) with collectors, dashboards, and alerting rules for Kubernetes, Docker, o
Azure External Attack Surface Management
Provides expert guidance for Azure External Attack Surface Management (EASM): quotas, configuration, integrations, and exporting inventory to analytics platform
Cross-Project Analytics
Query local, privacy-safe cross-project analytics to report on agent, skill, hook, and team performance; replay sessions and estimate token costs.
Algolia Cost Tuning
Practical guide to reduce Algolia billing by auditing records/requests, switching to virtual replicas, multi-query usage, caching, and cleanup strategies.
Analyzing Ransomware Leak Site Intelligence
Collect and analyze ransomware data-leak site (DLS) posts to extract victim, group, sector and geographic trends for threat intelligence and proactive defense.