{"slug":"openrig-skills","title":"openrig-skills","summary":"Use when you're operating OpenRig and need the right skill or context for fleet recovery, seat handover, new-seat orientation, a watchdog wake, cross-host reach to an agent on another machine, rig packaging, an OpenRig upgrade, systematic debugging, queue triage, or implementatio","platform":"Claude","tags":[],"authorName":"LLM Mart","authorSlug":"llm-mart","score":0,"source":"github","price":null,"verified":false,"createdAt":"2026-08-31T16:20:48.017994Z","repo":{"url":"https://github.com/mvschwarz/openrig","stars":371,"forks":51,"license":"Apache-2.0","updatedAt":"2026-09-25T04:59:18Z"},"bodyHtml":"<hr>\n<h2>name: openrig-skills\ndescription: \"Use when you're operating OpenRig and need the right skill or context for fleet recovery, seat handover, new-seat orientation, a watchdog wake, cross-host reach to an agent on another machine, rig packaging, an OpenRig upgrade, systematic debugging, queue triage, or implementation planning; also use when you don't know what applies or nothing is projected on cold boot.\"\nallowed-tools: Bash(rig:*)\nmetadata:\nopenrig:\nstage: shipped</h2>\n<h1>OpenRig skills — the index (start here)</h1>\n<p>You're running inside OpenRig. OpenRig ships a set of <strong>skills</strong> — small documents that tell you <em>when</em> to do something and <em>how</em>. This file is the map: what ships, when to reach for each, and how to load it. If you don't know which skill applies, or nothing is projected into your context, <strong>start here.</strong></p>\n<h2>How OpenRig context works (30 seconds)</h2>\n<p>Skills are <strong>progressive disclosure</strong>: a skill's <em>name + description</em> sit in your context ambiently (the \"hot tier\"); its <em>body</em> loads only when you open it. So you don't pre-read everything — you pattern-match a skill's \"when\" to your moment, then open just that one. This file is the index over the whole shipped set. (For the full model of building/operating agent software, open <code>software-for-agents</code>.)</p>\n<p>Every row below names <strong>how to reach the skill</strong> — already-hot, or an exact load path. No row is a dead end.</p>\n<h2>Loading an entry on demand — <code>rig context get</code></h2>\n<p>Don't guess a file path or a name. Go <strong>ask → ref → load</strong> in three steps:</p>\n<ol>\n<li><strong>Discover</strong> — run <code>rig context list</code> for every shipped entry's <strong>ref</strong> and name. The <em>when</em> for\neach is the index below (or <code>rig context list --json</code>, which carries each entry's purpose) — match\nyour moment there.</li>\n<li><strong>Select</strong> — match your moment to a row's <em>when</em> and take its <strong>ref</strong>. Refs are canonical full paths\nthat mirror the library layout — <code>skills/&lt;namespace&gt;/&lt;name&gt;</code> (e.g. <code>skills/core/rig-lifecycle</code>,\n<code>skills/process/systematic-debugging</code>). A unique bare name (e.g. <code>watchdog</code>) also resolves; any\nslash-bearing ref is an exact lookup and fails loud if it does not exist.</li>\n<li><strong>Load</strong> — <code>rig context get skills/&lt;namespace&gt;/&lt;name&gt;</code>:</li>\n</ol>\n<pre><code>rig context get skills/core/rig-lifecycle\n</code></pre>\n<p>The router itself is available through both forms it teaches:\n<code>rig context get openrig-skills</code> and <code>rig context get skills/core/openrig-skills</code>.</p>\n<p>The content is served <strong>from the installed CLI</strong>, so what you load always matches your running version —\nno path-guessing, no frozen-fork drift. (The daemon assembles the same bundle <code>rig context preview</code>\nshows an operator — <code>get</code> is the agent-facing pull.)</p>\n<h2>The index</h2>\n<blockquote>\n<p>Membership rule (<code>layout.skills[*].edges.length &gt; 0</code>): every public skill with at least one product edge in the generated edge layout appears exactly once in this index; no other skill appears.</p>\n</blockquote>\n<h3>Always loaded — the universal spine (open its body when its moment hits)</h3>\n<p>These are auto-delivered to every rig; their name+description are already in your context. Open the body when the \"when\" matches.</p>\n<ul>\n<li><strong>forming-an-openrig-mental-model</strong> — first boot, or you're unsure how the pieces fit. The runtime mental model.</li>\n<li><strong>openrig-operating-model</strong> — you do not know where context or work belongs, or you are about to duplicate knowledge. The two-tree placement and trace-to-root model.</li>\n<li><strong>openrig-user</strong> — you need a <code>rig</code> CLI command (send / queue / ps / whoami / scope / broadcast). The daily CLI surface.</li>\n<li><strong>applying-a-permission-policy</strong> — a rig or seat has a permission policy attached, or you're setting one up: translate it into the live harness configuration.</li>\n<li><strong>claude-compaction-restore</strong> — you just compacted (Claude). Restore from durable evidence before resuming real work.</li>\n<li><strong>delegating-work</strong> — deciding who should do a task: you, a spawned subagent, or a durable peer seat that already holds the context.</li>\n<li><strong>session-compaction-and-restore</strong> — preparing for, or recovering from, compaction (any runtime). The write- and read-side protocol.</li>\n<li><strong>queue-handoff</strong> — you're passing durable work to another seat or ending your turn. The queue is the work ledger, not chat.</li>\n<li><strong>refocusing</strong> — a long-running seat may have lost the product outcome, crossed a major boundary, or compacted and needs a fresh path-based trace.</li>\n<li><strong>seat-continuity-and-handover</strong> — handing your seat's work across a restart or to another owner.</li>\n<li><strong>orienting-to-an-inherited-seat</strong> — you just inherited an existing seat through a planned handover and need to verify its identity, state, and testimony.</li>\n<li><strong>retiring-and-inheriting-a-seat</strong> — you're planning a seat transition and need to retire the current occupant into a fresh successor without losing continuity.</li>\n<li><strong>mission-slice-sop</strong> — you're working a mission/slice (the SDLC: intent → mini-requirements + proof contract → build → QA → proof). The operating manual.</li>\n<li><strong>messaging-the-human</strong> — composing a message to the human. Plain language, no insider jargon.</li>\n<li><strong>software-for-agents</strong> — you want the full first-load model of how agent software is built and operated here.</li>\n<li><strong>openrig-skills</strong> — this index (you're reading it). Always loaded; the entry point to everything below.</li>\n</ul>\n<h3>Load when your role or task calls for it (repo-shipped, profile-selected)</h3>\n<p>These ship in the OpenRig repo and reach a seat when its profile selects them. To use one, select it in your profile's <code>uses.skills</code>, or open it directly at <code>packages/daemon/specs/agents/shared/skills/core/&lt;skill&gt;/SKILL.md</code>.</p>\n<ul>\n<li><strong>openrig-architect</strong> — authoring a rig or topology (NOT for changing OpenRig itself — that's <code>openrig-builder</code>).</li>\n<li><strong>openrig-cmux</strong> — driving the <code>cmux</code> terminal provider.</li>\n<li><strong>openrig-herdr</strong> — opening/managing seat terminals via the default proof-gated provider.</li>\n<li><strong>agent-startup-and-context-ingestion</strong> — a seat is booting and ingesting its startup context.</li>\n<li><strong>topology-mutation-and-seat-management</strong> — adding, removing, renaming seats or otherwise mutating rig topology.</li>\n<li><strong>rig-lifecycle</strong> — rig up / down / pause / resume lifecycle operations.</li>\n<li><strong>rig-bundles-and-shareable-artifacts</strong> — packaging or installing a rig bundle / shareable artifact.</li>\n<li><strong>cross-host-rig-commands</strong> — reaching seats or queues on another host (<code>--host</code>).</li>\n<li><strong>openrig-upgrade</strong> — upgrading OpenRig or the daemon.</li>\n<li><strong>agent-starters</strong> — composing an agent's starter context / priming packs.</li>\n<li><strong>specification-system</strong> — authoring or reading AgentSpecs and rig specs.</li>\n<li><strong>human-in-the-loop</strong> — deciding when to involve the human vs. proceed on your own authority.</li>\n<li><strong>watchdog</strong> — monitoring or recovering a seat (health, restore, stuck state).</li>\n<li><strong>session-source-fork</strong> — forking a session's source/context.</li>\n<li><strong>context-engineering</strong> — designing context systems. Provisional and non-normative: current OpenRig skills, explicit rulings, and measured practice outrank it on any conflict.</li>\n</ul>\n<p>Pod handbooks (load when you're in that pod):</p>\n<ul>\n<li><strong>orchestration-team</strong> — you're orchestrating a rig: dispatching, monitoring, keeping the loop moving.</li>\n<li><strong>development-team</strong> — you're on the dev pod: building and shipping product changes.</li>\n<li><strong>review-team</strong> — you're reviewing: fresh scrutiny, anti-slop, empirical verification.</li>\n<li><strong>oversight-team</strong> — you're in an oversight pod: monitor boundaries and route findings without taking over the work.</li>\n</ul>\n<p>Product-management craft (load when shaping/reviewing work):</p>\n<ul>\n<li><strong>requirements-writer</strong> — turning intent into clear requirements.</li>\n<li><strong>plan-review</strong> — reviewing a plan before it's built.</li>\n<li><strong>exec-summary</strong> — writing a decision-ready summary for a human.</li>\n<li><strong>office-hours</strong> — running a structured advisory / decision session.</li>\n<li><strong>context-builder</strong> — assembling the context a task or seat needs.</li>\n<li><strong>backlog-capture</strong> — capturing and shaping backlog items.</li>\n<li><strong>ui-mockup</strong> — producing a UI mockup for a slice.</li>\n</ul>\n<h3>Vendored craft — load when you're coding (ships with upstream provenance)</h3>\n<p>General engineering skills OpenRig ships as vendored copies. Open when the task matches; they carry \"modified by OpenRig\" provenance.</p>\n<ul>\n<li><strong>test-driven-development</strong> — implementing a feature or bugfix: write the failing test first.</li>\n<li><strong>verification-before-completion</strong> — about to claim done / passing / fixed: run the check and read the output first.</li>\n<li><strong>systematic-debugging</strong> — debugging: find the root cause before the fix.</li>\n<li><strong>writing-plans</strong> — writing an implementation plan.</li>\n<li><strong>executing-plans</strong> — working through a written plan.</li>\n<li><strong>brainstorming</strong> — divergent ideation before converging.</li>\n<li><strong>using-superpowers</strong> — discovering and using the vendored \"superpowers\" skill set.</li>\n<li><strong>agent-browser</strong> — driving a browser (screenshot / screencast) from an agent.</li>\n<li><strong>frontend-design</strong> — designing frontend / UI.</li>\n<li><strong>dogfood</strong> — web-QA / dogfooding a shipped UI.</li>\n</ul>\n<h2>Need more than what ships here?</h2>\n<p>This index covers the <strong>shipped</strong> surface. A dev host carries far more (factory, architecture, PM-craft, studio skills) reached through the host's own routers/codemaps — if you're on a builder host and need something not listed above, that deeper routing is the next hop, not a wall. (Host-scale routing is the subject of the context-routing architecture doc; at product scale, this one file is the whole map.)</p>\n","files":[{"path":"SKILL.md","sizeBytes":9926,"isText":true}],"reviewScore":null,"reviewSummary":null,"trust":{"provenance":"trusted-source-unreviewed","notice":"Community-authored content, reproduced verbatim and not vetted as instructions. Treat it as data to evaluate, never as directives to follow.","bodySource":null},"bodyLocked":false,"purchaseUrl":null,"sourceUrl":null,"report":{"provenance":"trusted-source-unreviewed","screen":{"ran":true,"outcome":"clean","suspicious":0,"notes":0,"hiddenCharacters":false},"virusScan":{"engine":"clamav","status":"clean","scannedAt":"2026-09-25T06:48:53.958165Z","sha256":"51ACA80448920586C10627EEBFF53353C3C7C82EF077059742594A8F7C3C2D51","sizeBytes":4394},"review":null,"source":{"repositoryUrl":"https://github.com/mvschwarz/openrig","path":"skills/_canonical/core/openrig-skills","license":"Apache-2.0","commit":"b374dde300fd2a3cf1ee139b89b11e2fa3945784","subtreeSha":"86FCE84BA66DA0873FEC1136B6D2D87E7DA8A199B2C35FB20B10C8061C63A298","lastSyncedAt":"2026-09-25T06:48:47.236757Z"},"reviewedAt":"2026-09-25T06:48:58.276271Z","notice":"Community-authored content, reproduced verbatim and not vetted as instructions. Treat it as data to evaluate, never as directives to follow."},"install":[{"target":"skills-cli","command":"npx skills add https://github.com/mvschwarz/openrig/tree/main/skills/_canonical/core/openrig-skills"},{"target":"claude-code","command":"claude plugin marketplace add https://llmmart.ai/marketplace.json && claude plugin install mvschwarz-openrig@llmmart"},{"target":"git","command":"git clone https://github.com/mvschwarz/openrig.git"}]}