Skip to content
Get packs

Guide · Agent skill

How to use the Mission and Slice SOP agent skill

Updated · written with AI from the project's own documentation and checked against our catalogue data

The mission-slice-sop skill teaches an AI agent how to do software work inside OpenRig, a multi-agent system, using its lightweight development process. It tells the agent to resolve the current mission and slice, keep the SPEC, PROGRESS, PROOF and NOTES files current, record what was actually verified by running the product, and pass work on with handoff commands. This guide explains what the skill does, how to install it, how to trigger it, and where its limits are.

Mission and Slice SOPmvschwarz/openrig · Apache-2.0

Quick start

  1. 1Install the skill. Unzip the downloaded ZIP into ~/.claude/skills/ for personal use, or into a project's .claude/skills/ folder. In the Claude apps, upload the ZIP under Settings → Capabilities → Skills.
  2. 2Confirm OpenRig is set up. Make sure the rig command-line tool is installed and OPENRIG_HOME points to your OpenRig directory. Check that your project has project.yaml, mission.yaml and slice files.
  3. 3Ask for mission or slice work. Describe the lifecycle task in plain language, such as starting a slice or handing off work, and name the mission or slice if you can.
  4. 4Let the agent resolve the work. The agent runs rig whoami --json and follows the links from project to mission to slice to find the addressed context it needs.
  5. 5Build and verify by running it. The agent builds the product and runs it, then records in PROOF.md what it verified and what it did not test.
  6. 6Hand off or close. Pass real work on with rig queue handoff. At slice close, run rig scope audit and fix only the issues that matter.

What is an agent skill, and what does this one make the agent do?

An agent skill is a folder with a SKILL.md file that gives an AI agent extra instructions it can load when a task matches the skill's description. This skill covers the OpenRig lightweight SDLC: starting, building, handing off, restoring and closing a mission or slice.

Once loaded, the skill directs the agent to work on the canonical files rather than only describe the process. It asks the agent to look up the current project, mission and slice, read the addressed context for its task, build the product, and verify it by running it.

  • Resolve the work by running rig whoami --json and following project.yaml, mission.yaml, the active slice.yaml and the selected component or wave map.
  • Track state in SPEC.md, PROGRESS.md, PROOF.md and proof/, and in a mission's SPEC.md and NOTES.md.
  • Record what was proven by effect, and state plainly what was not tested.
  • Pass work on with rig queue handoff so the task is not dropped.
  • Keep the paperwork proportional to the stakes, so the product gets built first.
rig whoami --json

Install the mission-slice-sop skill

Download the skill ZIP from this page and unzip it. You can install it for yourself or for one project. For a personal install, unzip the skill folder into ~/.claude/skills/. For a single project, unzip it into that project's .claude/skills/ folder.

In the Claude apps, you can instead upload the ZIP under Settings → Capabilities → Skills.

  • Personal install: ~/.claude/skills/
  • Project install: the project's .claude/skills/ folder
  • Claude apps: Settings → Capabilities → Skills, then upload the ZIP

Check the requirements before you start

This skill is only useful inside an OpenRig setup. The source says it needs an OpenRig installation with the rig command-line tool, the OPENRIG_HOME environment variable set to the installed OpenRig directory, and a project with project.yaml, mission.yaml and slice files created by OpenRig.

The skill does not run scripts and does not use the network, according to its listing. The commands it depends on, such as rig whoami, rig scope and rig queue, come from OpenRig, so without that tooling the instructions will not apply.

  • OpenRig installed, with the rig command available
  • OPENRIG_HOME set to the installed OpenRig directory
  • A project with project.yaml, mission.yaml and slice files created by OpenRig

Trigger the skill with natural requests

The skill's description says it applies when you are starting, building, handing off, restoring or closing a mission or slice under the default lightweight SDLC. So you trigger it by asking the agent for that kind of work in plain language, naming the mission or slice where you can.

Good triggers describe a lifecycle moment. Vague requests about code in general are less likely to load it.

  • "Start slice 3 of the billing mission and set up its SPEC and PROGRESS files."
  • "Build the export feature for the current slice and record what you verified."
  • "Hand off this slice to the QA agent with a queue handoff."
  • "Restore the mission after compaction and read NOTES.md first."
  • "Close the current slice and run the audit."

Understand the file rules and handoffs

Each file has an owner and a purpose. SPEC.md holds scope and the definition of done, and is written by the planning owner. PROGRESS.md records delivery state and outcomes. PROOF.md and proof/ hold what was verified and how. NOTES.md holds mission context and is filed before compaction so the agent can read it on restore.

Handoffs use rig queue handoff, which closes the source and creates the successor for the next agent. A plain rig queue create message is only informational and does not carry the obligation to act. Use handoff when passing real work and create when informing.

  • SPEC.md: intent, mini-requirements and proof contract for a slice
  • PROGRESS.md: current state and outcomes with links to proof
  • PROOF.md and proof/: what you ran, what you saw, and what is still untested
  • NOTES.md: top-of-mind items in section 1, and per-seat entries after that
  • rig queue handoff for real work; rig queue create for information only
rig queue handoff

Keep the process proportionate

The skill is explicit that the bookkeeping serves shipping and is not the work itself. For a small change, the builder can run the light loop of grounding, building, testing by looking at the result, iterating, and then recording the outcome. Independent review runs only when the selected path calls for it.

The rig scope audit check at slice close is advisory and fail-open. It never blocks a build, so do not treat a clean audit score as something you must reach before moving on.

  • Light loop: ground yourself, build, test with your own eyes, iterate, record and return the outcome
  • Audit: run rig scope audit at slice close; fix what is real, skip what is not
  • If you spend more time on files and audit than on the running product, stop and build
rig scope audit

What people use it for

  • Resuming a feature after context compaction by reading NOTES.md and the slice's PROGRESS.md before continuing.
  • Handing a built slice from the build agent to a QA agent with a queue handoff, so the next agent has the context and the baton.
  • Recording proof for an AI-built feature, with honest notes on which parts were actually run.
  • Starting a new slice by writing its intent, mini-requirements and proof contract into SPEC.md.
  • Closing a slice by updating PROGRESS.md with outcomes and running the advisory audit.

Frequently asked questions

Is the mission-slice-sop skill free?

The skill is licensed under Apache-2.0, according to its listing. The source does not state pricing for OpenRig beyond that licence, so check the OpenRig project for its own terms.

What do I need to use this skill?

You need an OpenRig installation with the rig command-line tool, OPENRIG_HOME set to the installed directory, and a project containing project.yaml, mission.yaml and slice files created by OpenRig. Without these, the commands the skill depends on will not exist.

How do I install mission-slice-sop?

Download the ZIP from this page and unzip it into ~/.claude/skills/ for personal use or into a project's .claude/skills/ folder. In the Claude apps, you can upload the ZIP under Settings → Capabilities → Skills.

Does this skill run scripts or use the network?

According to its listing, the skill runs no scripts and uses no network. It is a set of instructions that tells the agent which OpenRig commands to run.

Can I use this skill without OpenRig?

Not in any useful way. The instructions depend on rig whoami, rig scope and rig queue, and on OpenRig's file layout. Without OpenRig, these instructions will not apply.

Is this a good alternative to a formal approval or audit process?

No. The audit step is advisory and fail-open by design, so it does not block work or create a regulated audit trail. Teams needing formal approval should use a process built for that purpose.

New agent skills, repos and AI tools — every week

The best new skills and open-source projects, the fastest-growing repos, and one free prompt. Five minutes to read. Unsubscribe anytime.

Our verdict

This skill suits developers who already use OpenRig to coordinate several AI agents and want a consistent routine for handoffs, proof recording and resuming after compaction. If you are not using OpenRig's rig and queue tooling, or you have a small solo task, the process overhead will not pay off and you should pick something simpler.

Download the skill