Skip to content
Get packs

Agent Starters

Create, inspect and manage named per-seat starting points for AI agent roles, with provenance tracking.

Source Download skill
Repository
mvschwarz/openrig
Stars
8,553
Licence
Apache-2.0
Files
1 file

What it does

This skill teaches an agent how to create, refresh, inspect, promote and deprecate a reusable 'starter': a saved starting point for an AI agent seat that combines its role, startup context, an optional native session source and provenance notes. It defines a registry YAML format and a six-state lifecycle, and sets rules such as refusing to copy credentials, refusing to compact a seat just to make a starter smaller, and reporting staleness honestly. It is explicitly not a VM-style image that captures machine state.

Great for

  • Developers building OpenRig rigs who want to reuse a known-good agent seat configuration
  • Teams maintaining a registry of agent role starters and needing provenance visibility
  • Anyone auditing whether a starter's source and context are traceable and still current

Skip it if

  • Users wanting deterministic VM-style state capture; the skill states starters do not capture VM state
  • Anyone wanting to copy provider API keys or auth material into a starter; the skill forbids this, so it would not work as intended
  • Tools outside the OpenRig daemon, since the referenced commands and resolver belong to that system and the file alone cannot run them

Try asking

  • “Create a new reviewer starter from a seat that has proven capable at code review”
  • “Inspect this starter and tell me its provenance and whether it is still recommended”
  • “Write a registry entry for a starter named reviewer-v1 with a role and context”
  • “Deprecate an old starter whose source transcript path is no longer valid”
Step-by-step guide: how to use Agent StartersInstall, first run, examples, tips and FAQs

Install in 3 steps

  1. 1

    Download the skill

    The ZIP contains the 1 original files, the Apache-2.0 licence and an attribution note.

    agent-starters.zip
  2. 2

    Add it to your agent

    Claude Code (all projects)

    mkdir -p ~/.claude/skills && unzip agent-starters.zip -d ~/.claude/skills/

    Claude Code / Agent SDK (this project only)

    mkdir -p .claude/skills && unzip agent-starters.zip -d .claude/skills/

    Claude apps (web/desktop): Settings → Capabilities → Skills → Upload skill, and choose the ZIP.

  3. 3

    Ask for the task

    No special command needed — the agent reads the skill's description and loads it when your request matches. Try one of the prompts above.

Get better results

  • Record provenance (manifest id/version, runtime, source session id, ready-check evidence, freshness) when a starter is based on a primed session, because promotion is refused without inspectable provenance.
  • Note that a RigSpec member combining starter_ref with session_source.mode fork is currently rejected, so plan native continuity via a separate session-source path.
  • Do not shrink a seat to make the starter smaller; the skill treats size as an outcome, not a target.

What's inside

  • SKILL.md

Redistributed unmodified from mvschwarz/openrig under the Apache-2.0 licence. The assessment was drafted with AI from the skill's files on 2026-10-29. Skills can instruct your agent to run code — read SKILL.md and any scripts before installing.