/preflight
Check provider health before running multi-LLM workflows
> /plugin marketplace add nyldn/claude-octopus
Command definition
preflight.mdcommand: preflight
disable-model-invocation: true
description: Check provider health before running multi-LLM workflows
allowed-tools: Bash
Octopus Provider Readiness
Show the shared readiness result for every registered provider. The default is a local-only static check: installation, safe configuration metadata, and known quota state. It does not contact provider services.
OCTO_ROOT="${CLAUDE_PLUGIN_ROOT:-${CODEX_PLUGIN_ROOT:-${HOME}/.claude-octopus/plugin}}"
bash "${OCTO_ROOT}/scripts/helpers/preflight.sh"When the user explicitly asks to verify live provider health, run the bounded live form:
bash "${OCTO_ROOT}/scripts/helpers/preflight.sh" --liveBoth modes consume the Provider Registry 2.0 readiness contract. Do not run separate binary, authentication, model, quota, or network checks while rendering the result. Use `reason_code` and `remediation` from the shared objects when explaining a provider that needs attention.
When `OCTOPUS_COUNCIL_DEFAULT_PROVIDERS` pins a council roster (for example `claude,agy` when Codex leads the session), preflight checks exactly those seats, names the list it used, and reports a seat that is not ready; the orchestrator's `preflight` command fails in that case, and so does its smoke test when any roster seat it tested fails. A roster seat with no provider detection, such as `openai-tools`, is listed as not checked (`roster_unchecked` in the JSON) rather than as not ready. When preflight runs inside a Codex or Claude Code sandbox, a provider that needs a keychain login, the network or files outside the workspace can fail there and work outside it; the output says so, and the remedy is to re-run preflight outside the sandbox.
Use this before `/octo:embrace`, `/octo:research`, or `/octo:council` to confirm provider readiness and avoid mid-workflow surprises.
**Common issues:**
- `Ollama` requests a live check → rerun with `--live`; start it with `ollama serve` if the server is stopped
- `Codex CLI` unavailable → run `codex login` or `npm install -g @openai/codex`
- `Antigravity CLI` unavailable → verify `agy` is on PATH, then launch plain `agy` and complete its browser sign-in or reinstall Antigravity CLI
Run `/octo:setup` to install or configure any missing provider.
Read more
command: preflight disable-model-invocation: true description: Check provider health before running multi-LLM workflows allowed-tools: Bash
Octopus Provider Readiness
Show the shared readiness result for every registered provider. The default is a local-only static check: installation, safe configuration metadata, and known quota state. It does not contact provider services.
OCTO_ROOT="${CLAUDE_PLUGIN_ROOT:-${CODEX_PLUGIN_ROOT:-${HOME}/.claude-octopus/plugin}}"
bash "${OCTO_ROOT}/scripts/helpers/preflight.sh"When the user explicitly asks to verify live provider health, run the bounded live form:
bash "${OCTO_ROOT}/scripts/helpers/preflight.sh" --liveBoth modes consume the Provider Registry 2.0 readiness contract. Do not run separate binary, authentication, model, quota, or network checks while rendering the result. Use `reason_code` and `remediation` from the shared objects when explaining a provider that needs attention.
When `OCTOPUS_COUNCIL_DEFAULT_PROVIDERS` pins a council roster (for example `claude,agy` when Codex leads the session), preflight checks exactly those seats, names the list it used, and reports a seat that is not ready; the orchestrator's `preflight` command fails in that case, and so does its smoke test when any roster seat it tested fails. A roster seat with no provider detection, such as `openai-tools`, is listed as not checked (`roster_unchecked` in the JSON) rather than as not ready. When preflight runs inside a Codex or Claude Code sandbox, a provider that needs a keychain login, the network or files outside the workspace can fail there and work outside it; the output says so, and the remedy is to re-run preflight outside the sandbox.
Use this before `/octo:embrace`, `/octo:research`, or `/octo:council` to confirm provider readiness and avoid mid-workflow surprises.
**Common issues:**
- `Ollama` requests a live check → rerun with `--live`; start it with `ollama serve` if the server is stopped
- `Codex CLI` unavailable → run `codex login` or `npm install -g @openai/codex`
- `Antigravity CLI` unavailable → verify `agy` is on PATH, then launch plain `agy` and complete its browser sign-in or reinstall Antigravity CLI
Run `/octo:setup` to install or configure any missing provider.
Every AI model has blind spots. Claude Octopus supports twelve external provider integrations — Codex, Antigravity CLI, Copilot, Qwen, Ollama, Perplexity, OpenRouter, OrcaRouter, OpenCode, Cursor CLI, Grok, and Kimi Code — alongside the built-in Claude Code
Other commands on octo.
More commands from other plugins.
setup
CommandOne-time setup after installing ARIS as a Claude Code plugin — registers the Codex reviewer MCP server from the plugin directory…
autoresearch
CommandAutonomous iteration loop: modify, verify, keep/discard against any metric
agent
Commandname: sc:agent description: SC Agent — session controller that orchestrates investigation, implementation, and review category:…

