analyze
CommandAnalyze recent runs - per-model latency, tokens, and verdict agreement - and suggest model/reasoning/fanout tuning. Advisory,…
Ask a single configured OpenRouter model for a second opinion. Advisory only. Single-shot or multi-turn.
> /plugin marketplace add antonbabenko/deliberation> /plugin install deliberation@antonbabenko-deliberation
name: ask-openrouter description: Ask a single configured OpenRouter model for a second opinion. Advisory only. Single-shot or multi-turn. allowed-tools: mcp__deliberation__ask-openrouter, mcp__deliberation-openrouter__openrouter, mcp__deliberation-openrouter__openrouter-reply, mcp__deliberation-openrouter__openrouter-list, Read, Bash timeout: 660000
Delegate a question to the default OpenRouter model or a specific configured delegate (advisory only; OpenRouter cannot edit files).
`$ARGUMENTS` is `[alias] <question>`. If the first whitespace-delimited token matches a configured delegate alias, it selects that delegate and the rest is the question. Otherwise the whole string is the question and the default OpenRouter model configured in `config.json` (via `models.<id>.default: true` or `providers.openrouter.defaultModel` / `model`) is called.
1. Identify the expert role from the question via `~/.claude/rules/deliberation/triggers.md` (default Architect). Then load that expert's prompt: 1. Glob `~/.claude/plugins/cache/*/deliberation/*/prompts/[expert].md` and pick the match with the highest semver version segment (the segment immediately after `deliberation/`, parsed as semver - not lexical compare). 2. If no match is found, abort with: `Error: deliberation plugin cache missing for expert "[Expert]". Run /plugin install deliberation or /reload-plugins.` 2. Build the 7-section delegation prompt per `~/.claude/rules/deliberation/delegation-format.md`. Time-sensitive facts: when the question turns on latest or current versions, pricing, a roadmap, or whether a model or tool exists, verify it with your own tools first and add the facts with as-of date and source (the delegate cannot look anything up). If the question references local files, attach them with `files: [{ path: "...", mode: "auto" }]` (text-inline; `{ dir: "..." }` also supported). 3. Print: `OpenRouter (<alias-or-default>) working (typical 30-60s)...` where `<alias-or-default>` is the selected alias, or `default` when no alias was given. 4. Call OpenRouter:
mcp__deliberation__ask-openrouter({
prompt: "[7-section prompt]",
developerInstructions: "[expert prompt]",
cwd: "[repo root]",
preset: "[optional preset id, e.g. cheapest-first]",
files: [ /* optional text-inline files */ ]
}) mcp__deliberation-openrouter__openrouter({
prompt: "[7-section prompt]",
"developer-instructions": "[expert prompt]",
alias: "[selected alias, or omit to call default]",
cwd: "[repo root]",
files: [ /* optional text-inline files */ ]
})5. On `result.isError`, report the `errorKind` (`model-not-allowed` => bad alias; `auth` => the env var named by `apiKeyEnv` is empty; `config` => a hard config failure, fix `config.json`). A single bad model entry no longer breaks the whole config: this single-shot call still works as long as the requested alias or default model is valid. 6. Synthesize the answer; never paste raw output. For a follow-up turn, reuse the returned `threadId` via `mcp__deliberation-openrouter__openrouter-reply`.
Codex/Gemini.
Get a second opinion from GPT, Gemini, and Grok in Claude Code, Codex, Cursor, Kiro, Antigravity, OpenCode, or any other MCP host - plus local models (Ollama, LM Studio) and 400+ more models through OpenRouter, including Qwen, Kimi, and DeepSeek.
Analyze recent runs - per-model latency, tokens, and verdict agreement - and suggest model/reasoning/fanout tuning. Advisory,…
Ask GPT, Gemini, Grok, and any configured OpenRouter models in parallel for independent second opinions, then synthesize and…
Get Gemini second opinion on a question or current work. Single-shot, advisory, no contamination. Model pinned per call.
Get GPT (Codex) second opinion on a question or current work. Single-shot, advisory, no contamination.
Get Grok (xAI) second opinion on a question or current work. Single-shot, advisory, no contamination.
Log GPT (Codex) in on this machine with a ChatGPT device login - shows a link and a one-time code. For remote or headless…
Implement tasks incrementally — build, test, verify, commit. Add "auto" to run the whole plan in one approved pass.