Skill UI

Provides comprehensive patterns and techniques for securing AI agent systems. Learn how to enforce governance policies, including tool allow/block lists, content filtering, rate limiting, and human approval workflows. Essential for building robust, compliant, and trustworthy agents in multi-agent or production environments that interact with external APIs and sensitive data.

Implement robust human-in-the-loop (HITL) approval gates for sophisticated LangGraph workflows. This skill enables the graph to pause execution before critical, irreversible actions (like sending emails or calling tools). It manages complex state serialization, provides clear resume semantics using `Command(resume=...)`, and integrates with external UIs (e.g., Slack) to await human decision, ensuring safe and controlled workflow execution.

Implements a comprehensive defense-in-depth framework for autonomous AI agents that interact with external tools. It mitigates risks like prompt injection or tool misuse by enforcing strict tool allowlisting, binding tools to least-privilege identities, requiring human-in-the-loop approval for high-impact actions, and utilizing policy enforcement (like NeMo Guardrails) to bound the blast radius and ensure compliance.

This skill structures a highly controlled, bounded development life cycle for AI agents. It mandates a structured three-phase workflow: Planning (Spec), Implementation (Build), and Verification (Review). It is essential for building complex features from scratch, especially when dealing with ambiguous requirements, high risk, or needing explicit human approval gates to ensure the resulting code meets all defined specifications.

This skill manages repeated rollouts and recursive prompting by maintaining an append-only decision ledger. It tracks trials, marks, and coherence checks to prevent auto-approval of live trading or destructive actions. Ideal for high-dimensional search, stochastic optimization, and ensemble comparison requiring visible evidence trails.

This meta-skill automates the entire short-drama production pipeline, starting from a simple topic. The workflow analyzes the request to infer render style, character identity, and shot count. It first drafts a detailed, shot-by-shot shooting script and pauses for user review. After approval, it generates a universal cast reference image, per-shot composition images, and corresponding video clips. Crucially, it maintains character and scene consistency by anchoring every video to both the reference image and the composition image. Finally, it adds title/ending cards, burns subtitles, and de...

This skill enforces strict communication protocols for AI agents interacting in shared channels with external parties. It ensures agents speak only when explicitly addressed, prevents internal context leakage, and routes sensitive content to draft-only approval. Ideal for group chats, DMs, and business partnerships where privacy and security are critical.

Defines an operator approval contract for agent-drafted outbound messages: hashed draft filing, epoch-keyed decisions, approval snapshots, durable delivery claims and receipts, plus a pre-draft baseline gate. Use when a human must approve, reject, or steer each send before it leaves.

This skill structures a highly controlled, bounded development life cycle for AI agents. It mandates a structured three-phase workflow: Planning (Spec), Implementation (Build), and Verification (Review). It is essential for building complex features from scratch, especially when dealing with ambiguous requirements, high risk, or needing explicit human approval gates to ensure the resulting code meets all defined specifications.

Official Beatra AI agent package for paid text-to-video, image-to-video generation, and video editing/extension services. Produces short AI video clips on hosted Beatra service with first/last-frame interpolation and reference-guided generation capabilities. Requires explicit user approval and accepts paid processing on third-party hosted infrastructure.

Runs the Hermes Agent CLI in one-shot mode (hermes -z) to delegate bounded local coding tasks from Code Mate. Covers availability checks, timeouts, exit codes, authentication limits and YOLO tool-approval safety constraints.

Configure approved delegation lanes across installed implementer CLIs, with optional model and effort choices. Discovers available CLIs, proposes a lane map, and writes global or project config only after explicit user approval.