Agent skills for Devin Desktop

Coordinates structured peer reviews and iterative feedback loops between AI agents like Claude and Codex.

Design autonomous agent systems and the human oversight around them — specifying what an agent can perceive and act on, why a stateless…

Interact with Discord servers - send messages, read channels, manage reactions

Interact with Discord servers using bot tokens - send messages, read channels, manage reactions

Use when rendering multi-agent debate, discussion, or review output in the terminal.

Multi-agent orchestration enforcement for /team command.

Use when summarizing agent evaluations where autonomous, assisted, failed, timed-out, or invalid outcomes must remain distinct and…

Designs and runs agent evaluation frameworks to measure and improve performance.

Patterns and techniques for adding governance, safety, and trust controls to AI agent systems.

Optimizes and generates AI coding agent instruction files like AGENTS.md and CLAUDE.md to improve project context and model performance.

Use when creating, editing, or deploying custom agent instructions or skills shared by Codex and Claude.

Use when writing or editing instructions for agents, including AGENTS.md, CLAUDE.md, SKILL.md, project rules, or reusable agent prompts.

Structured self-debugging workflow for AI agent failures using capture, diagnosis, contained recovery, and introspection reports.

Use when planning or implementing on this owner's projects.

Your long-term knowledge pages. Read them at session start. Create new pages when you learn something worth remembering across sessions…

Stores and manages persistent user preferences, decisions, and cross-session context in structured markdown memory files.

Send meme reaction images in chat. One command to pick & send. Multi-platform (Discord, Feishu, Telegram, etc). Use when the conversation…

Use when deciding when and how to commit items to a project memory system.

Operate explicit orchestrator, implementer, validator, and scribe roles through a caller-selected agent runtime.

Reviews code to ensure agent-native parity — any action a user can take, an agent can also take.

Read-only observation of an agent's parent, siblings, and direct children.