team-avatar Plugin

• @agent-team-avatar:aang

  Keine Beschreibung in der Datei gefunden.

• @agent-team-avatar:appa

  Keine Beschreibung in der Datei gefunden.

• @agent-team-avatar:iroh

  Keine Beschreibung in der Datei gefunden.

• @agent-team-avatar:katara

  Keine Beschreibung in der Datei gefunden.

• @agent-team-avatar:momo

  Keine Beschreibung in der Datei gefunden.

• @agent-team-avatar:sokka

  Keine Beschreibung in der Datei gefunden.

• @agent-team-avatar:suki

  Keine Beschreibung in der Datei gefunden.

• @agent-team-avatar:toph

  Keine Beschreibung in der Datei gefunden.

• @agent-team-avatar:zuko

  Keine Beschreibung in der Datei gefunden.

• agni-kai

  Zuko's Agni Kai for bakeoffs. The formal firebending duel, where competing implementations fight under the same rules and a judge declares the winner. Use for arena, /arena, "arena this", "throw it in the arena", "bakeoff", or when one attempt at a nontrivial design or artifact would lock in the wrong shape. Spawns N parallel candidates at the same task, picks a base with a cross-judge, and grafts the strongest parts of the losers into it.

• avatar-help

  Explains the Team Avatar roster, skill invocation methods for Claude Code and Codex, and provides guidance on selecting the right skill for a task.

• avatar-mode

  Routes complex engineering tasks to specialized agent personas based on discipline, such as architecture, debugging, or performance.

• avatar-setup

  Configures model assignments, reasoning effort, and session hook routing for Team Avatar characters in avatar-mode.

• battle-map

  Maintains a TSV-based decision log to provide an audit trail of actions, reasoning, and evidence for long or unattended agent runs.

• break-camp

  Resumes paused or handed-off agent work by reading checkpoints, transcripts, or branches to rebuild state and continue from the last known point.

• breath-of-fire

  Optimizes software performance by capturing baselines, identifying bottlenecks, and applying a prioritized hierarchy of fixes to meet specific metric targets.

• checkup

  Performs periodic audits of verification skills and feature maps to identify documentation drift and product regressions.

• clear-skies

  Resolves git merge conflicts non-interactively, regenerates lockfiles, and verifies builds and tests before staging changes.

• clear-water

  Cleans code by running a comment reviewer pass, fixing root causes of code surprises, and replacing explanatory comments with types, tests, or lints.

• day-of-black-sun

  Designs bespoke, auditable workflows for ambitious, multi-part changes or large migrations using a hypothesis-driven scientific method.

• dragon-dance

  Implements a scientific hillclimbing loop to iteratively improve a single metric through hypothesis testing and measurement.

• earth-rumble

  Performs parallel correctness and quality reviews on code diffs, branches, or PRs to identify bugs, security holes, and maintainability issues.

• ember-trace

  Performs runtime forensics on system artifacts like cpuprofiles, traces, and heap snapshots to provide cited diagnoses of performance issues.

• four-elements

  Provides 24 engineering principles categorized into four elements to guide decision-making during refactoring, debugging, and architectural design.

• full-patrol

  Automates the end-to-end lifecycle of independent pull requests from build to merge-ready status using parallel subagents and verification lanes.

• heal-ci

  Automatically identifies and fixes failing CI checks by analyzing logs and applying iterative, minimal code corrections.

• healing-water

  Reproduces defects, binary-searches for root causes using runtime evidence, and implements minimal fixes verified on the original failure surface.

• honor-code

  Prevents repeated agent mistakes by implementing architectural, type-based, or linting safeguards based on historical error patterns.

• iceberg-wake

  Summarizes chat changes and action items using plain language and visuals to catch a user up after they have been away.

• invention-workshop

  Creates or updates personalized agent mode skills by mining user interaction history and preferences to automate working styles.

• kyoshi-command

  Orchestrates long-running, multi-day projects by managing hundreds of subagents through a structured coordinator-worker hierarchy and a persistent state store.

• kyoshi-formation

  Orchestrates parallel agent workers to perform swarm tasks, coverage sweeps, or races and aggregates their results into a single report.

• lie-detector

  Performs multi-model adversarial reviews by dispatching diverse AI models to stress test code and identify blind spots.

• light-load

  Improves pull request reviewability by cleaning commit history, sharpening descriptions, and adding reviewer guidance without changing code behavior.

• long-flight

  Automates the management of pull requests by resolving conflicts, addressing review comments, and monitoring CI status until they are merge-ready.

• make-camp

  Creates a durable work checkpoint by committing progress and publishing a resume note for cold-start agents to resume tasks.

• master-plan

  Creates structured, verifiable plans for multi-step features, migrations, or multi-PR programs by breaking work into units with specific checks.

• meditation

  Analyzes conversation transcripts using parallel reviewers to extract durable learnings and automatically propose updates to agent skills.

• messenger-hawk

  Fetches and categorizes pull request review comments into an ordered action list based on severity and actionability.

• metalbending

  Performs strict code quality audits to identify structural impurities, spaghetti logic, and maintainability issues while pushing for dramatic simplification.

• my-cabbages

  Analyzes code changes to identify potential blast radius and breakage by proving safety assumptions through real code execution.

• new-scroll

  Author and modify agent skills using a structured format with frontmatter, progressive disclosure, references, and scripts.

• night-watch

  Executes a single long-running task autonomously by driving toward a defined exit condition without human intervention.

• pai-sho

  Tests agent strategies and prompt changes by running blinded candidates against a judge using organic user prompts and transcript verification.

• past-lives

  Queries multiple evidence sources in parallel to provide cited, confidence-tiered explanations for design decisions and code history.

• plain-talk

  Restates complex messages in simple, jargon-free language while preserving all original facts and numbers.

• purify

  Removes AI-generated code slop from git diffs by eliminating unnecessary comments, abnormal defensive checks, and deep nesting without changing behavior.

• push-and-pull

  Implements a test-driven development rhythm of pushing failing tests and pulling code to satisfy intended behavior.

• redemption-arc

  Manages production incidents and postmortems by stabilizing systems, building blameless timelines, and converting lessons into structural fixes.

+18

• hooks

  Keine Beschreibung in der Datei gefunden.