Resolve a capability

A pstack step names what it needs by a capability name in backticks, such as spawn-worker. It does not name a tool. This skill says which tool does it here. The skills that name capabilities live next to this one, so read this file as ../resolve-capability/SKILL.md.

1. Which environment is this

Answer these as separate questions. They are not alternatives. Orca runs on top of a harness, so you can be in Claude Code and in Orca at the same time.

Which harness are you? Claude Code, Codex, or Grok. Your own tool list tells you.
Is Orca here? Run env | grep '^ORCA_CLI_' now, before you plan or start anything, and also when the task says to plan only. Any output means yes. Do not decide this from the harness you are, from which, or from memory, and do not put the check off to a later step. An Orca terminal sets only ORCA_CLI_BIN_DIR and an Orca chat sets both. Run Orca commands with $ORCA_CLI_COMMAND when it is set, otherwise with orca. Below, orca means that executable. Orca adds workers, durable run state, schedules, remote environments, and session search. No output means a plain session: read references/plain.md.
This computer and this user. ~/.agents/skills/select-model/SKILL.md records which products and models are installed, which Orca environments exist, and what the user prefers. Read it too. When it is missing, say so once and use the defaults in ../poteto-mode/SKILL.md.

Read the file for your harness once, and the Orca files when Orca is here: claude-code.md, codex.md, grok.md.

2. How to use the index

Each line below gives the mechanism per environment and then what to do when there is none. A mechanism marked (unverified) has not been run on a real session yet. Use it, and say in your reply if it failed.