Agent skills, MCP & prompting

Structured self-debugging workflow for AI agents to diagnose and recover from failures with actionable reports.

Investigate what went wrong in a Superpowers AI agent session by reading transcripts and producing evidence-backed findings for bug reports.

Breaks work into ordered tasks. Use when you have a spec or clear requirements and need to break work into implementable tasks. Use when a task feels

A workflow that executes coding plans by delegating each task to a fresh subagent, reviewing each task's output, and finishing with a whole-branch rev

Determines the appropriate skill or workflow for a given task based on the development phase.

A router skill that asks about your situation and recommends which skill or flow in the collection fits best.

Coordinates multiple specialized agents to work independently on separate problems in parallel, then integrates their results.

A meta-skill that establishes how an AI agent discovers and invokes other skills at the start of every conversation before taking any action.

Analyzes a repository's actual stack to sort agent skills, commands, rules, and hooks into always-loaded (DAILY) versus searchable (LIBRARY) buckets,

Helps you set up and manage the context agents receive—rules files, specs, relevant files, and context budgeting—so agent output stays accurate and fo

Build and manage voice AI agents with natural conversations and custom tools

Scaffolds agent company packages conforming to the agentcompanies/v1 spec, including agents, teams, workflows, skills, and Paperclip config.

Researches, authors, or references a new skill and publishes it into the Paperclip skills catalog with a validated manifest and pull request.

A meta-skill that guides agents through creating, editing, and verifying other agent skills using a test-driven development approach with pressure sce

A skill that helps an agent ask good questions and understand what a user actually wants before building or planning anything.

Provides patterns for orchestrating multiple parallel AI agent sessions using dmux, a tmux-based pane manager for agent harnesses like Claude Code and

A meta-skill that forces the AI to produce complete, unabridged outputs with no placeholders, truncation, or skipped sections.

Displays a one-shot reference card explaining the Ponytail lazy-coding plugin's levels, skills, slash commands, and configuration options.

Guides building and qualifying external bot integrations (like Dot in Paperclip) covering invitations, identity, Runner lifecycle, native tools, recov

Ponytail-help provides a quick reference for ponytail's modes, skills, and commands in a one-shot display

Recommends matching courses, tutorials, and use cases from Claude Academy when users ask how to use Claude or its features.

Runs an hourly health-check and observation-sync cycle for a local claude-mem worker, maintaining a deduplicated middle cache, exclude marks, and a ru

Searches and reads local coding-agent session logs across tools like Codex, Claude Code, and OpenCode to recover what past sessions or subagent runs d

Generates a HANDOFF.md briefing so a fresh Claude session can resume long or degraded sessions with full context of goals, failures, and next steps.

Searches and retrieves relevant information from persistent cross-session memory.

Customize what Claude-Mem remembers by creating and managing custom modes.

Orchestrates a named team of cooperating Codex agents with durable on-disk state, choosing between native MultiAgentV2 agents and Codex App threads, a

Explains any senpi/omo tip in depth by querying the live tip list and reading the actual feature code, tailored to which tips the user can personally

Orchestrates a five-member adversarial agent team that cross-critiques a planning request, distills only the surviving insights, and hands them to a d

Runs a hostile multi-agent debate to stress-test a plan before formalizing it into an executable work plan.

Generates a hierarchical AGENTS.md knowledge base across a repository so AI agents can quickly understand its structure, commands, and conventions.

Orchestrates dependency-ordered multi-agent work by defining and driving DAGs of tasks through the native workflow tool, with retry, amend, and send r

Configures the claude-mem-cowork plugin by collecting and storing the user's cmem.ai credentials (sync token, user ID, SyncHub URL) so memory capture

Walks a new omo user through a guided first-session tour, setup migration, session analysis, and memory recording.

Searches the user's persistent Claude-Mem memory to recall past sessions, decisions, and observations about projects or tasks.

A placeholder template skill used as a starting point for creating new agent skills.