Analyzes a failed superpowers session and drafts a bug report for the maintainers.

Dispatches independent tasks to parallel agents, each working without shared state or sequential dependencies.

Lets the agent implement a written plan inline in the current session, without subagents.

Guides integrating finished, fully tested work by choosing merge, PR, or cleanup options.

Executes independent plan tasks through subagents within the current session.

Teaches the agent to find and invoke relevant skills before responding to any message.

Guides creating, editing, and testing skills before they are deployed to agents.

Routes you to the right mattpocock skill or flow for your current situation.

Pursue a long-running goal in one session by coordinating multiple subagents.

Hand the current conversation to a fresh background agent that continues the work immediately.

Compact the current conversation into a handoff document for another agent to resume.

Interview me about specs for workflows I want to build in this workspace.

Configure a repo's issue tracker, triage labels, and domain docs for engineering skills.

Turn an unanswerable decision into a questionnaire for someone else to complete.

Split a plan or conversation into tracer-bullet tickets with blocking edges, published to a tracker.

Make the agent re-pitch a message that failed to land clearly.

Map large multi-session work as decision tickets and resolve them until the path is clear.

Audit a 12-layer LLM agent stack for regressions, memory pollution, tool misuse, and hidden loops.

Design agent action spaces, tool definitions, and observation formats to improve task completion rates.

Diagnose AI agent failures with structured self-debugging, containment, recovery, and introspection reports.

Trim ECC to a repo's needed skills by sorting them into daily and library buckets.

Operate as an agentic engineer with eval-first execution, task decomposition, and cost-aware model routing.

Build a persistent multi-agent operating system in Claude Code with specialists, commands, memory, and scheduling.

AIエージェントが大量の実装を生成するチームの開発とレビューの運用を設計する。

ジョブやフック、MCPサーバーの稼働状況と冗長性を証拠に基づいて監査する。

Claude Codeに永続メモリやスケジュール実行を加え、常時稼働の自動エージェントを構築する。

単純なパイプラインからマルチエージェントDAGまで、自動Claude Codeループを設計する。

複数セッションのプロジェクトを、自己完結したステップ付きの構築計画に変換する。

Persist project memory across Claude Code sessions, auto-loading context and tracking git activity deterministically.

Orchestrate parallel coding agents in isolated worktrees, monitoring their progress and reading structured reports.

Audit Claude Code context window usage and get prioritized recommendations for cutting token waste.

Extract reusable skills from sessions via a legacy Stop hook; deprecated in favor of v2.

Build atomic instincts from session hooks, scoped per project to avoid cross-project contamination.

Run a complete verification system for Claude Code sessions to check work before finishing.

Prompt a discernment check after substantive advice or drafts so users verify claims before acting on them.

Build MCP servers in Python or TypeScript that expose external APIs as well-designed LLM tools.

Create, evaluate, benchmark, and optimize skills, including tuning their descriptions for triggering.

Writes a quality bar into CONSTRAINTS.md and flags agent changes that quietly weaken it.

Configures agent context and rules files to keep output quality high across sessions.

Cross-examines decisions with fresh-context adversarial review to surface hidden failure modes.

Discovers and selects the right agent skill for a session or task at hand.

Submit an unstructured complaint about agent tooling proactively, without asking permission or previewing.

Hands user-requested work to an existing Paperclip agent with a durable task reference.

Hands user-requested work to an existing Paperclip agent with a durable task reference.

Hands user-requested work to an existing Paperclip agent with a durable task reference.

Guide a user's first Paperclip task, proposing a plan and waiting for approval before hiring or executing.

Finds earlier Paperclip deliverables and attaches user feedback with human attribution.

Finds earlier Paperclip deliverables and attaches user feedback with human attribution.

Finds earlier Paperclip deliverables and attaches user feedback with human attribution.

Call the Paperclip control plane API to check assignments, update issues, post comments, and delegate work.

Call Paperclip control plane API to check assignments, update issues, post comments, and delegate.

Read Paperclip task context, deliver documents, report blockers, and follow governance during heartbeats.

Manage a Paperclip company through chat, covering onboarding, agents, approvals, costs, and work review.

Decide whether plan work needs delegation and link justified tasks to owners and real dependencies.

Hire new Paperclip agents by comparing configs, drafting prompts, and submitting governance-aware hire requests.

Creates, updates, comments on, and lists Paperclip tasks from Hermes using scoped credentials.

Research and prepare MCP vendor integrations with cited findings, approved content PRs, and governed connector PRs.

Turn an agent's execution record into evidence-backed, minimal instruction or tool-description change proposals.

Summarizes a connected Paperclip team's agents, tasks, blockers, and pending approvals needing attention.

Summarizes a connected Paperclip team's agents, tasks, blockers, and pending approvals needing attention.

Summarizes a connected Paperclip team's agents, tasks, blockers, and pending approvals needing attention.

Read Slack conversations and collaborate through the assigned Slack bot during tasks and routines.

Quietly submit a concrete improvement suggestion after noticing repeated, generalizable friction in agent work.

Monitor a pull request until reviews and CI issues are all resolved.

Run the hourly CCS Align cycle, syncing needle observations into a seat-owned middle cache.

Execute a phased implementation plan step by step through subagents.

Run phased plans through subagents, especially those created by make-plan.

Lets a Grok Bot answer claude-mem observer jobs with only skip or observation XML.

Create a phased implementation plan with documentation discovery before executing it.

Create a detailed phased implementation plan with documentation discovery first.

Configures claude-mem-cowork plugin credentials to pair cmem.ai cloud sync.

Create and verify custom claude-mem modes with domain observation types and tags.

Connects agents to 21risk via Rube MCP, looking up current tool schemas first.

Connects agents to Agent Mail via Rube MCP, looking up current tool schemas first.

Drive Anchor Browser automation tasks through Rube MCP, checking current tool schemas first.

Finds and reconstructs past coding-agent sessions and subagent runs from local transcript logs.

Interpret and respond to automatic comment-checker feedback emitted after edit-style PostToolUse hooks.

Saves a multi-agent DAG under a name so it can be re-run without pasting it again.

Runs an adversarial team that cross-critiques a plan before it gets formalized.

File source-backed bug issues or PRs for LazyCodex or Codex CLI defects with reproduction steps.

Runs dependency-ordered child tasks in phases, with retry, amend, and send recovery.

Explains Codex rules behavior, where rule files live, how they match, and how to configure them.

Creates and coordinates named teams of Codex workers with durable, script-managed state.

Applies the binding ultrawork-mode directive so the agent works in ultrawork mode.

Executes written ulw plans with state tracking, evidence logs, worktree discipline, and parallel subagents.

Runs a goal-like loop that breaks work into evidence-bound steps with durable checkpoints.

Decomposes goals into systematic, evidence-bound ultrawork steps in a checkpointed loop.

Produces one decision-complete work plan before coding, only when the user explicitly requests it.

Writes one explore-first, decision-complete plan before implementation starts on request.

Writes one decision-complete work plan through explore-first consultation before any coding begins.

Set up recurring, self-running marketing loops that an AI agent executes on a schedule or trigger.

Run hypothesis tree experiments with persistent evidence tracking across repeatable model and benchmark evaluators.

Detects repeated research workflows in Screenpipe history and stages new skill drafts for review.