Coordinate work across AI agents with shared rules for locating team files, recording decisions and their reasons, and requesting help. It also defines who owns revisions after a reviewer rejects an artifact.
Set up an AI agent team by proposing roles and names for a project, then creating shared team files after confirmation. Initialize member context, decision records, and logs so the team starts with a common project background.
Guides agents to record confirmed final outcomes in history.md and correct entries when decisions change. This keeps handoff records clear enough for future agents to understand what actually happened without checking other files.
Set up personal AI agents that join Squad sessions across projects. Manage their roles in a user configuration directory, inspect the combined team, and use agents that read project state and offer advice without writing to it.
Prepare architecture proposals before implementation so the team can agree on a direction. Organize evidence, technical choices, effects on existing work, risks, and the first release’s scope into a document with clear recommendations and decision owners.
Capture interface or documentation pages with Playwright and embed the screenshots in a GitHub pull request description. Reviewers can inspect visual changes without checking out the branch, using images hosted on a temporary branch.
Keep tests aligned with API changes and actual project files. This skill guides agents to update related tests in the same commit, refresh expected counts and lists, and check stale assertions when automated checks fail.
Find and select Squad operations through a categorized menu, from managing an AI team and GitHub issues to choosing models and reviewing activity. The skill collects required options and routes requests to CLI commands or the coordinator agent.
Manage work and personal GitHub identities in agent workflows by checking the active account and selecting credentials for each operation. It covers personal repository pushes, pull requests from forks, and separate configuration directories for account isolation.
Learn how to start an AI team with Squad in Copilot CLI. This reference explains how to invoke the coordinator agent, find the command catalog, and initialize a new project while avoiding common invocation mistakes.
Guide agents through failures with timed retries, alternative methods, targeted fixes, and clear handoffs. It also explains when to continue with partial results, helping teams choose a recovery approach that fits the problem.
Guides Squad documentation writing and review using Microsoft Style Guide rules and project conventions. It covers clear English, consistent page structure, practical examples near the top, and matching test updates when documentation pages are added.
Adapt Squad agent coordination to the available Copilot interface. Detect supported tools, choose how to launch helper agents and collect results, and use file-based state for workflows that must work across interfaces.
Plan task handoffs between agents on different machines using Git to share YAML task and result files. This implementation specification describes validation, execution limits, result reporting, and GitHub Issues as an additional route for urgent requests.
Find and start common Squad operations through a categorized menu. The coordinator guides users through team management, GitHub issues, plugins, model settings, and session history, collecting required details and confirming changes where the catalog requires it.
Clean up stored agent state in `.squad/` by compressing histories, pruning old logs, archiving stale decisions, and removing orphaned inbox files. Use it after long sessions or before starting work with many agents to reclaim context space.
Apply Squad SDK and CLI versioning rules to keep package versions aligned and development-only versions out of shared branches. It explains how prerelease versions affect npm workspace dependencies and defines release ownership and CI checks.
Find interrupted Copilot CLI sessions in stored history, review conversations, checkpoints, files, and linked issues or pull requests, then resume unfinished work with the context needed to understand where it stopped.
Follow Squad's Git workflow from issue branches through pull requests and cleanup. It covers development from dev, isolated directories for parallel issues, coordinated changes across repositories, and promotion to preview, insiders, and stable releases.
Manage rejected work in an AI team by assigning revisions to a different agent. Defines author lockouts, independent revision requirements, specialist escalation, and when the coordinator must ask the user to resolve a deadlock.
Extract lessons from corrections, praise, and unexpected cases in conversations. Organize findings by confidence level and propose updates to individual agent histories or shared team decisions, with changes shown for approval before saving.
Review agent charters and histories, extract repeated knowledge into shared skills, and shorten the original documents while preserving unique responsibilities. A before-and-after size report shows how much context each agent saves.
Set up separate commands for personal and work GitHub accounts on Windows. Detects accounts already signed in through GitHub CLI, asks which serves each role, creates PowerShell shortcuts and CMD wrappers, and verifies the selected usernames.
Review technical claims and deliverables by checking evidence and testing alternative explanations. Confirm that references and API endpoints exist, then report verified, unverified, or contradicted claims alongside evidence and an overall review verdict.
Is this your account? Link GitHub to bring it onto your creator page (Japanese)