Install 11 ready-to-use skills from whyashthakker/video-intelligence so Claude Code, Cursor, Codex, Gemini CLI, and VS Code can connect to Ceptory MCP, watch video recordings, configure 10 live camera workflows, triage alerts, set up RTSP cameras, and generate operational reports out of the box.
Skills give your agent Ceptory playbooks; the MCP server or REST API gives it live workspace access.
1
Install the skills
Run npx skills add whyashthakker/video-intelligence in your terminal (or copy any folder from skills/ into .agents/skills/ or .claude/skills/).
2
Connect Ceptory MCP or API
Add https://mcp.ceptory.com/mcp to your agent for browser OAuth sign-in, or use a workspace API key against https://www.ceptory.com/api/v1.
3
Delegate real video ops
Ask your agent to set up camera workflows, triage alerts with visual proof, trace movement across sites, configure RTSP edge boxes, or draft incident reports.
11 skills included
Built directly from Ceptory's platform
Every skill is grounded in live Ceptory endpoints, MCP tools, workflow schemas, and operational report templates.
01 · Video Search & Reportswatch-video
Watch & Inspect Video With Any LLM
Let any LLM watch local or remote videos: upload or link a recording, read its chronological visual + audio timeline, jump to exact moments, ask visual questions, and zoom into key frames.
Best for: Asking an LLM to watch a screen recording, bug repro, CCTV clip, product demo, or drone flight without blowing context limits.
Connect Claude Code, Claude Desktop, Cursor, VS Code, Codex, Gemini CLI, Windsurf, OpenAI Agents SDK, LangChain, or n8n to Ceptory MCP over OAuth 2.1 or the stdio local-file bridge.
Best for: Setting up Ceptory MCP, checking workspace balance and RBAC roles, or troubleshooting 401/402/429 errors.
Index recorded videos from HTTPS URLs or local files up to 2 GB, search across scenes, speech, and on-screen text, and run grounded video Q&A or REST API v1 pipelines.
Best for: Uploading footage, polling indexing status, running semantic queries, or integrating /api/v1/indexes.
Translate plain-English monitoring requests into 10 live camera workflows (PPE, person down, fire & hazards, intrusion, face access, loitering, unattended bags, crowd limit, illegal parking, rider helmets).
Best for: Creating or tuning camera rules, IANA schedules, normalized ROI zones, and clip/push/alarm responses.
Check camera fleet health, paginate open alerts, visually inspect annotated alert photos before resolving false alarms, and route notifications to Slack, Teams, Discord, Email, or HMAC-signed webhooks.
Best for: Running shift health checks, clearing false alarms safely, or verifying X-Ceptory-Signature alert webhooks.
Manage staff, watchlist, and custom allow/alert people lists on face access cameras, plus extract plates from photos and manage Authorized or Banned plate lists on parking cameras.
Best for: Enrolling staff or watchlists, tuning matchThreshold, reading plates from photos, or blocking banned vehicles.
Group cameras into physical sites, trace person and vehicle journeys across cameras with sighting thumbnails, geocode site addresses, and configure heading/FOV for automatic multi-camera GPS triangulation.
Best for: Following where a person or vehicle moved across a site or pinpointing overlapping alerts on a map.
Discover and run individual AI detection engines (PPE, fire/smoke, plates, faces, crowd, vehicles) on single images or stateful multi-frame tracking loops with optional annotated JPEG output.
Best for: Testing computer-vision detectors on still frames or building custom frame-by-frame pipelines on /api/v1/engines.
Connect existing IP cameras, NVRs, DVRs, and analog CCTV over outbound HTTPS using local motion gating and RTSP sub-stream templates for 15+ brands including Hikvision, Dahua, CP Plus, Axis, Reolink, and UniFi Protect.
Best for: Generating edge.toml, discovering ONVIF streams, fixing RTSP errors, or calling /api/v1/edge.
Follows the open Agent Skills standard (SKILL.md frontmatter + structured playbooks). Install with the CLI or drop folders directly into your project.
Claude Code
Cursor
Codex
Gemini CLI
VS Code
Windsurf
Antigravity
Any Agent Skills client
Install all 11 skills
Adds the full Ceptory video intelligence skill pack from GitHub in one command.
npx skills add whyashthakker/video-intelligence
Pair with Ceptory MCP
Give your agent live access to cameras, alerts, movement, video search, and AI engines.
claude mcp add --transport http ceptory https://mcp.ceptory.com/mcp
FAQ
Agent skills, answered
What are Ceptory Agent Skills?
They are 11 modular SKILL.md instruction packs following the open Agent Skills specification. When installed into your coding or operations agent, they teach it how to connect Ceptory MCP, watch videos without overflowing context, configure all 10 camera workflows, set up RTSP edge boxes, and draft operational reports accurately.
How do these skills work with the Ceptory MCP server?
The MCP server gives your agent live tools to call Ceptory, while these skills give the agent the playbooks for using those tools well—such as merging existing camera configs before updating, visually verifying alert photos before resolving false alarms, or formatting incident reports without guessing identities.
Can I install just one skill instead of all 11?
Yes. Run npx skills add whyashthakker/video-intelligence --skill <skill-name> to install a single skill, or copy any individual folder from skills/ on GitHub into your agent's .agents/skills/ or .claude/skills/ directory.
Which AI agents support these skills?
Any agent supporting the open Agent Skills format, including Claude Code, Cursor, Codex, Gemini CLI, VS Code, Windsurf, and Antigravity.
Are the skills open source?
Yes. The entire repository (github.com/whyashthakker/video-intelligence) is MIT-licensed. You can inspect every SKILL.md file, fork it for your organization's SOPs, or contribute improvements.
Open source + live tools
Pair these skills with the Ceptory MCP server
Install the 11 skills from GitHub, connect https://mcp.ceptory.com/mcp in two minutes, and run your cameras and video search from any AI agent.