@mvschwarz
Building OpenRig. Running rogue AI civilizations. Debugging their coordination problems.
55 skills1 repository 8K
Set up OpenRig to run a team of coding agents, including Claude Code and Codex. Open the team interface, pass project goals to its operator, reconnect to agents, and follow assigned work through to its results.
Opens groups of OpenRig agent terminals as live herdr tiles for interaction or observation. Explains view selection, read-only access, saved groups, and how to interpret agents that are offline or cannot be displayed.
Evaluate a feature idea before turning it into requirements using six sequential questions. Examine customer demand, current workarounds, a specific user's pain, and the smallest useful solution, then save an evidence-based assessment with a verdict and next steps.
Turn rough feature ideas into a clear SPEC.md through focused questions. Define target users, business rules, current scope, and acceptance criteria based on observable results, giving developers or AI agents concrete requirements to implement.
Create screen mockups at three levels: text wireframes, standalone HTML files, or working prototype pages. Use requirements, existing app styles, and real project data to show layouts and interactions for review or testing.
Coordinate work across an OpenRig agent team with clear assignments, attached context, completion criteria, and selected review checkpoints. Track queue status, identify blocked or idle owners, and resolve handoff gaps while leaving implementation with the assigned agents.
Review assigned code changes against project conventions, architecture, and intended behavior. Verify findings with source code and tests, classify their severity, and record a merge readiness verdict, with an optional deeper process for independent reviews and resolving disagreements.
Develop features and fix bugs by writing a behavior test first, observing its expected failure, and adding the minimum code needed to pass. Then clean up the code while keeping tests passing, applying this process where regression risk warrants it.
Manage OpenRig development work from setup to completion using shared specification, progress, and evidence files. Record what was actually checked and pass work to the next owner so handoffs and conversation compaction preserve the context.
Hand off OpenRig work through persistent queue items that record the next owner, blockers, and closure reasons. The skill covers notifying recipients, escalating to a human, and keeping supporting evidence accessible across interruptions and missed messages.
Look up exact syntax, options, JSON output, defaults, and error meanings for a known OpenRig command. Use the reference to check command behavior and diagnose runtime differences against the installed help and running daemon.
Plan and verify OpenRig startup, shutdown, and recovery. Compare reported status with actual running sessions, distinguish recovery outcomes, and define reboot tests that expose partial failures instead of hiding them behind a fresh launch.
Guides agents through sending human decision requests and updates using registered recipients. It covers concise briefs, delivery checks, duplicate prevention, and linking replies to the correct work so that delivery receipts are kept distinct from human decisions.
Turn rough meeting notes, customer requests, and product ideas into structured backlog entries, a list of possible improvements. Capture who needs each idea and why, categorize it, and update matching entries with new evidence.
Collects and summarizes background research for a product feature, including customer needs, competitors, regulations, and internal decisions. Produces a source-linked background.md and updates shared context documents when findings are useful across features.
Design and run bounded procedures in environments where current conditions are uncertain. The skill separates agent judgment from precise tool actions, verifies each effect, and records checkpoints so interrupted work can resume from observed evidence.
Organize knowledge and specifications in OpenRig so agents can find the right context and preserve what they learn. The skill explains where files belong and how to read documents from a local role up to broader shared guidance.
Help a long-running agent reconnect with the intended product outcome after context compaction or a major work transition. Trace the current work hierarchy and read relevant source files to rebuild grounding before the next consequential action.
Package an OpenRig agent arrangement and workflow for reuse by others. Guides bundle creation, inspection before installation, and checks in a clean environment to identify hidden dependencies on local paths, providers, or credentials.
Define reusable OpenRig agent setups, roles, and workflows in specification files. Review dependencies that prevent use on another host, distinguish specifications from bundles and extensions, and verify both file structure and actual execution.
Configure a new OpenRig agent seat, a managed place for an agent to run, using a prior Claude or Codex conversation. Covers fork settings, new session identities, and checks that distinguish forking from restoring.
Monitor other agent teams for abandoned work, process drift, off-task activity, and excessive token use. It guides evidence checks after alerts and chooses a proportionate response, from notifying the team’s coordinator to escalating to a human.
Find and load the appropriate skill for an agent working inside OpenRig. This index maps situations such as recovery, handover, debugging, and planning to relevant guidance, with commands that retrieve content matching the installed CLI.
Set up an OpenRig team for ongoing repository work, with clear ownership, implementation, independent review, and handoffs. Choose direct coordination, a task queue, or an explicit workflow, then expand the team as authorized work requires.
Is this your account? Link GitHub to bring it onto your creator page (Japanese)